1 dnl To generate the html version, execute
2 dnl m4 web/manual.m4 | grutatxt --toc
4 define(`LOCAL_LINK_NAME', `translit(`$1', `A-Z
6 define(`REMOVE_NEWLINE', `translit(`$1',`
9 define(`REFERENCE', ./``#''`LOCAL_LINK_NAME($1)' (`REMOVE_NEWLINE($2)'))
10 define(`XREFERENCE', `$1' (`REMOVE_NEWLINE($2)'))
11 define(`EMPH', ``_''`REMOVE_NEWLINE($1)'``_'')
16 This document describes how to install, configure and use the paraslash
17 network audio streaming system. Most chapters start with a chapter
18 overview and conclude with an example section. We try to focus on
19 general concepts and on the interaction of the various pieces of the
20 paraslash package. Hence this user manual is not meant as a replacement
21 for the manual pages that describe all command line options of each
28 In this chapter we give an REFERENCE(Overview, overview) of the
29 interactions of the two main programs contained in the paraslash
30 package, followed by REFERENCE(The paraslash executables, brief
31 descriptions) of all executables.
36 The core functionality of the para suite is provided by two main
37 executables, para_server and para_audiod. The former maintains a
38 database of audio files and streams these files to para_audiod which
39 receives and plays the stream.
41 In a typical setting, both para_server and para_audiod act as
42 background daemons whose functionality is controlled by client
43 programs: the para_audioc client controls para_audiod over a local
44 socket while the para_client program connects to para_server over a
45 local or remote networking connection.
47 Typically, these two daemons run on different hosts but a local setup
50 A simplified picture of a typical setup is as follows
53 server_host client_host
54 ~~~~~~~~~~~ ~~~~~~~~~~~
56 +-----------+ audio stream +-----------+
57 |para_server| -----------------------------> |para_audiod|
58 +-----------+ +-----------+
69 | connect +-----------+
70 +-------------------------------------- |para_client|
75 The paraslash executables
76 ~~~~~~~~~~~~~~~~~~~~~~~~~
80 para_server streams binary audio data (MP3, OGG/Vorbis, OGG/Speex,
81 M4A, WMA files) over local and/or remote networks. It listens on a
82 TCP port and accepts commands such as play, stop, pause, next from
83 authenticated clients. There are many more commands though, see the
84 man page of para_server for a description of all commands.
86 It supports three built-in network streaming protocols
87 (senders/receivers): HTTP, DCCP, or UDP. This is explained in more
88 detail in the section on REFERENCE(Networking, networking).
90 The built-in audio file selector of paraslash is used to manage your
91 audio files. It maintains statistics on the usage of all available
92 audio files such as last-played time, and the number of times each
95 Additional information may be added to the database to allow
96 fine-grained selection based on various properties of the audio file,
97 including information found in (ID3) tags. However, old-fashioned
98 playlists are also supported.
100 It is also possible to store images (album covers) and lyrics in the
101 database and associate these to the corresponding audio files.
103 The section on the REFERENCE(The audio file selector, audio file
104 selector) discusses this topic.
109 The client program to connect to para_server. paraslash commands
110 are sent to para_server and the response is dumped to STDOUT. This
111 can be used by any scripting language to produce user interfaces with
112 little programming effort.
114 All connections between para_server and para_client are encrypted
115 with a symmetric RC4 session key. For each user of paraslash you must
116 create a public/secret RSA key pair for authentication.
121 The local daemon that collects information from para_server.
123 It runs on the client side and connects to para_server. As soon as
124 para_server announces the availability of an audio stream, para_audiod
125 starts an appropriate receiver, any number of filters and a paraslash
126 writer to play the stream.
128 Moreover, para_audiod listens on a local socket and sends status
129 information about para_server and para_audiod to local clients on
130 request. Access via this local socket may be restricted by using Unix
131 socket credentials, if available.
136 The client program which talks to para_audiod. Used to control
137 para_audiod, to receive status info, or to grab the stream at any
138 point of the decoding process.
142 A command line HTTP/DCCP/UDP stream grabber. The http mode is
143 compatible with arbitrary HTTP streaming sources (e.g. icecast).
147 A filter program that reads from STDIN and writes to STDOUT.
148 Like para_recv, this is an atomic building block which can be used to
149 assemble higher-level audio receiving facilities. It combines several
150 different functionalities in one tool: decoders for multiple audio
151 formats and a number of processing filters, among these a normalizer
156 A small stand-alone program that prints tech info about the given
157 audio file to STDOUT. It can be instructed to print a "chunk table",
158 an array of offsets within the audio file or to write the content of
159 the audio file in complete chunks 'just in time'.
161 This allows third-party streaming software that is unaware of the
162 particular audio format to send complete frames in real time.
166 A modular audio stream writer. It supports a simple file writer
167 output plug-in and optional WAV/raw players for ALSA (Linux) and for
168 coreaudio (Mac OS). para_write can also be used as a stand-alone WAV
174 Curses-based gui that presents status information obtained in a curses
175 window. Appearance can be customized via themes. para_gui provides
176 key-bindings for the most common server commands and new key-bindings
182 An (OSS-only) alarm clock and volume-fader.
188 This chapter lists the REFERENCE(Requirements, necessary software)
189 that must be installed to compile the paraslash package, describes
190 how to REFERENCE(Installation, compile and install) the paraslash
191 source code and the steps that have to be performed in order to
192 REFERENCE(Quick start, set up) a typical server and client.
197 In any case you'll need
199 - XREFERENCE(http://systemlinux.org/~maan/osl/, libosl).
200 The _object storage layer_ library is used by para_server. To
201 clone the source code repository, execute
203 git clone git://git.tuebingen.mpg.de/osl
205 - XREFERENCE(ftp://ftp.gnu.org/pub/gnu/gcc, gcc). The
206 EMPH(gnu compiler collection) is usually shipped with the
207 distro. gcc-3.3 or newer is required.
209 - XREFERENCE(ftp://ftp.gnu.org/pub/gnu/make, gnu make) is
210 also shipped with the disto. On BSD systems the gnu make
211 executable is often called gmake.
213 - XREFERENCE(ftp://ftp.gnu.org/pub/gnu/bash, bash). Some
214 scripts which run during compilation require the EMPH(Bourne
215 again shell). It is most likely already installed.
217 - XREFERENCE(http://www.openssl.org/, openssl). The EMPH(Secure
218 Sockets Layer) library is needed for cryptographic routines
219 on both the server and the client side. It is usually shipped
220 with the distro, but you might have to install the "development
221 package" (called libssl-dev on debian systems) as well.
223 - XREFERENCE(ftp://ftp.gnu.org/pub/gnu/help2man, help2man)
224 is used to create the man pages.
228 - XREFERENCE(http://www.underbit.com/products/mad/, libmad).
229 To compile in MP3 support for paraslash, the development
230 package must be installed. It is called libmad0-dev on
231 debian-based systems. Note that libmad is not necessary on
232 the server side, i.e. for sending MP3 files.
234 - XREFERENCE(http://www.underbit.com/products/mad/,
235 libid3tag). For version-2 ID3 tag support, you'll need
236 the libid3tag development package libid3tag0-dev. Without
237 libid3tag, only version one tags are recognized.
239 - XREFERENCE(http://www.xiph.org/downloads/, ogg vorbis).
240 For ogg vorbis streams you'll need libogg, libvorbis,
241 libvorbisfile. The corresponding Debian packages are called
242 libogg-dev and libvorbis-dev.
244 - XREFERENCE(http://www.audiocoding.com/, libfaad). For aac
245 files (m4a) you'll need libfaad (libfaad-dev).
247 - XREFERENCE(http://www.speex.org/, speex). In order to stream
248 or decode speex files, libspeex (libspeex-dev) is required.
250 - XREFERENCE(ftp://ftp.alsa-project.org/pub/lib/, alsa-lib). On
251 Linux, you'll need to have ALSA's development package
252 libasound2-dev installed.
254 - XREFERENCE(http://downloads.xiph.org/releases/ao/,
255 libao). Needed to build the ao writer (ESD, PulseAudio,...).
256 Debian package: libao-dev.
261 First make sure all non-optional packages listed in the section on
262 REFERENCE(Requirements, required software) are installed on your
265 You don't need everything listed there. In particular, MP3, OGG/Vorbis,
266 OGG/Speex and AAC support are all optional. The configure script will
267 detect what is installed on your system and will only try to build
268 those executables that can be built with your setup.
270 Note that no special decoder library (not even the MP3 decoding library
271 libmad) is needed for para_server if you only want to stream MP3 or WMA
272 files. Also, it's fine to use para_server on a box without sound card.
274 Next, install the paraslash package on all machines, you'd like this
277 (./configure && make) > /dev/null
279 There should be no errors but probably some warnings about missing
280 packages which usually implies that not all audio formats will be
281 supported. If headers or libs are installed at unusual locations you
282 might need to tell the configure script where to find them. Try
286 to see a list of options. If the paraslash package was compiled
287 successfully, execute (optionally)
291 to run the paraslash test suite. If all tests pass, execute as root
295 to install executables under /usr/local/bin and the man pages under
301 *Step 1*: Create a paraslash user
303 In order to control para_server at runtime you must create a paraslash
304 user. As authentication is based on the RSA crypto system you'll have
305 to create an RSA key pair. If you already have a user and an RSA key
306 pair, you may skip this step.
308 In this section we'll assume a typical setup: You would like to run
309 para_server on some host called server_host as user foo, and you want
310 to connect to para_server from another machine called client_host as
313 As foo@server_host, create ~/.paraslash/server.users by typing the
317 target=~/.paraslash/server.users
318 key=~/.paraslash/id_rsa.pub.$user
319 perms=AFS_READ,AFS_WRITE,VSS_READ,VSS_WRITE
320 mkdir -p ~/.paraslash
321 echo "user $user $key $perms" >> $target
323 Next, change to the "bar" account on client_host and generate the
324 key pair with the commands
326 ssh-keygen -t rsa -b 2048
327 # hit enter twice to create a key with no passphrase
329 This generates the two files id_rsa and id_rsa.pub in ~/.ssh. Note
330 that paraslash can also read keys generated by the "openssl genrsa"
331 command. However, since keys created with ssh-keygen can also be used
332 for ssh, this method is recommended.
334 Note that para_server refuses to use a key if it is shorter than 2048
335 bits. In particular, the RSA keys of paraslash 0.3.x will not work
336 with version 0.4.x. Moreover, para_client refuses to use a (private)
337 key which is world-readable.
339 para_server only needs to know the public key of the key pair just
340 created. Copy this public key to server_host:
342 src=~/.ssh/id_rsa.pub
343 dest=.paraslash/id_rsa.pub.$LOGNAME
344 scp $src foo@server_host:$dest
346 Finally, tell para_client to connect to server_host:
348 conf=~/.paraslash/client.conf
349 echo 'hostname server_host' > $conf
352 *Step 2*: Start para_server
354 Before starting the server make sure you have write permissions to
355 the directory /var/paraslash that has been created during installation:
357 sudo chown $LOGNAME /var/paraslash
359 Alternatively, use the --afs_socket Option to specify a different
360 location for the AFS command socket.
362 For this first try, we'll use the info loglevel to make the output
363 of para_server more verbose.
367 Now you can use para_client to connect to the server and issue
368 commands. Open a new shell as bar@client_host and try
373 to retrieve the list of available commands and some server info.
374 Don't proceed if this doesn't work.
376 *Step 3*: Create and populate the database
378 An empty database is created with
382 This initializes a couple of empty tables under
383 ~/.paraslash/afs_database-0.4. You normally don't need to look at these
384 tables, but it's good to know that you can start from scratch with
386 rm -rf ~/.paraslash/afs_database-0.4
388 in case something went wrong.
390 Next, you need to add some audio files to that database so that
391 para_server knows about them. Choose an absolute path to a directory
392 containing some audio files and add them to the audio file table:
394 para_client add /my/mp3/dir
396 This might take a while, so it is a good idea to start with a directory
397 containing not too many files. Note that the table only contains data
398 about the audio files found, not the files themselves.
400 You may print the list of all known audio files with
404 *Step 4*: Configure para_audiod
406 para_audiod needs to create a "well-known" socket for the clients to
407 connect to. The default path for this socket is
409 /var/paraslash/audiod_socket.$HOSTNAME
411 In order to make this directory writable for para_audiod, execute
414 sudo chown $LOGNAME /var/paraslash
417 We will also have to tell para_audiod that it should receive the
418 audio stream from server_host via http:
420 para_audiod -l info -r '.:http -i server_host'
422 You should now be able to listen to the audio stream once para_server
423 starts streaming. To activate streaming, execute
427 Since no playlist has been specified yet, the "dummy" mode which
428 selects all known audio files is activated automatically. See the
429 section on the REFERENCE(The audio file selector, audio file selector)
430 for how to use playlists and moods to specify which files should be
431 streamed in which order.
435 It did not work? To find out why, try to receive, decode and play the
436 stream manually using para_recv, para_filter and para_write as follows.
438 For simplicity we assume that you're running Linux/ALSA and that only
439 MP3 files have been added to the database.
441 para_recv -r 'http -i server_host' > file.mp3
442 # (interrupt with CTRL+C after a few seconds)
443 ls -l file.mp3 # should not be empty
444 para_filter -f mp3dec -f wav < file.mp3 > file.wav
445 ls -l file.wav # should be much bigger than file.mp3
446 para_write -w alsa < file.wav
448 Double check what is logged by para_server and use the --loglevel
449 option of para_recv, para_filter and para_write to increase verbosity.
455 para_server uses a challenge-response mechanism to authenticate
456 requests from incoming connections, similar to ssh's public key
457 authentication method. Authenticated connections are encrypted using
458 the RC4 stream cipher.
460 In this chapter we briefly describe RSA and RC4 and sketch the
461 REFERENCE(Client-server authentication, authentication handshake)
462 between para_client and para_server. User management is discussed
463 in the section on REFERENCE(The user_list file, the user_list file).
464 These sections are all about communication between the client and the
465 server. Connecting para_audiod is a different matter and is described
466 in a REFERENCE(Connecting para_audiod, separate section).
473 RSA is an asymmetric block cipher which is used in many applications,
474 including ssh and gpg. An RSA key consists in fact of two keys,
475 called the public key and the private key. A message can be encrypted
476 with either key and only the counterpart of that key can decrypt
477 the message. While RSA can be used for both signing and encrypting
478 a message, paraslash only uses RSA only for the latter purpose. The
479 RSA public key encryption and signatures algorithms are defined in
482 RC4 is a stream cipher, i.e. the input is XORed with a pseudo-random
483 key stream to produce the output. Decryption uses the same function
484 calls as encryption. While RC4 supports variable key lengths,
485 paraslash uses a fixed length of 256 bits, which is considered a
486 strong encryption by today's standards. Since the same key must never
487 be used twice, a different, randomly-generated key is used for every
490 Client-server authentication
491 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
493 The authentication handshake between para_client and para_server goes
496 - para_client connects to para_server and sends an
497 authentication request for a user. It does so by connecting
498 to para_server, TCP 2990, the control port of para_server.
500 - para_server accepts the connection and forks a child process
501 which is supposed to handle the connection. The parent process
502 keeps listening on the control port while the child process
503 (also called para_server below) continues as follows.
505 - para_server loads the RSA public key of that user, fills a
506 fixed-length buffer with random bytes, encrypts that buffer
507 using the public key and sends the encrypted buffer to the
508 client. The first part of the buffer is the challenge which
509 is used for authentication while the second part is the RC4
512 - para_client receives the encrypted buffer and decrypts it
513 using the user's private key, thereby obtaining the challenge
514 buffer and the session key. It sends the SHA1 hash value of
515 the challenge back to para_server and stores the session key
518 - para_server also computes the SHA1 hash of the challenge
519 and compares it against what was sent back by the client.
521 - If the two hashes do not match, the authentication has
522 failed and para_server closes the connection.
524 - Otherwise the user is considered authenticated and the client
525 is allowed to proceed by sending a command to be executed. From
526 this point on the communication is encrypted using the RC4
527 stream cipher with the session key known to both peers.
529 paraslash relies on the quality of openssl's cryptographically strong
530 pseudo-random bytes, on the security of the implementation of the
531 openssl RSA and RC4 crypto routines and on the infeasibility to invert
534 Neither para_server or para_client create RSA keys on their own. This
535 has to be done once for each user as sketched in REFERENCE(Quick start,
536 Quick start) and discussed in more detail REFERENCE(The user_list
542 At startup para_server reads the user list file which must contain
543 one line per user. The default location of the user list file may be
544 changed with the --user_list option.
546 There should be at least one user in this file. Each user must have
547 an RSA key pair. The public part of the key is needed by para_server
548 while the private key is needed by para_client. Each line of the
549 user list file must be of the form
551 user <username> <key> <perms>
553 where _username_ is an arbitrary string (usually the user's login
554 name), _key_ is the full path to that user's public RSA key, and
555 _perms_ is a comma-separated list of zero or more of the following
558 +---------------------------------------------------------+
559 | AFS_READ | read the contents of the databases |
560 +-----------+---------------------------------------------+
561 | AFS_WRITE | change database contents |
562 +-----------+---------------------------------------------+
563 | VSS_READ | obtain information about the current stream |
564 +-----------+---------------------------------------------+
565 | VSS_WRITE | change the current stream |
566 +---------------------------------------------------------+
568 The permission bits specify which commands the user is allowed to
569 execute. The output of
573 contains in the third column the permissions needed to execute the
576 It is possible to make para_server reread the user_list file by
577 executing the paraslash "hup" command or by sending SIGHUP to the
581 Connecting para_audiod
582 ~~~~~~~~~~~~~~~~~~~~~~
584 para_audiod listens on a Unix domain socket. Those sockets are
585 for local communication only, so only local users can connect to
586 para_audiod. The default is to let any user connect but this can be
587 restricted on platforms that support UNIX socket credentials which
588 allow para_audiod to obtain the Unix credentials of the connecting
591 Use para_audiod's --user_allow option to allow connections only for
592 a limited set of users.
594 -----------------------
595 The audio file selector
596 -----------------------
598 paraslash comes with a sophisticated audio file selector (AFS),
599 whose main task is to determine which file to stream next, based on
600 information on the audio files stored in a database. It communicates
601 also with para_client whenever an AFS command is executed, for example
602 to answer a database query.
604 Besides the traditional playlists, AFS supports audio file selection
605 based on _moods_ which act as a filter that limits the set of all
606 known audio files to those which satisfy certain criteria. It also
607 maintains tables containing images (e.g. album cover art) and lyrics
608 that can be associated with one or more audio files.
610 AFS uses libosl, the object storage layer, as the backend library
611 for storing information on audio files, playlists, etc. This library
612 offers functionality similar to a relational database, but is much
613 more lightweight than a full database backend.
615 In this chapter we sketch the setup of the REFERENCE(The AFS process,
616 AFS process) during server startup and proceed with the description
617 of the REFERENCE(Database layout, layout) of the various database
618 tables. The section on REFERENCE(Playlists and moods, playlists
619 and moods) explains these two audio file selection mechanisms
620 in detail and contains pratical examples. The way REFERENCE(File
621 renames and content changes, file renames and content changes) are
622 detected is discussed briefly before the REFERENCE(Troubleshooting,
623 Troubleshooting) section which concludes the chapter.
628 On startup, para_server forks to create the AFS process which opens
629 the OSL database tables. The server process communicates with the
630 AFS process via pipes and shared memory. Usually, the AFS process
631 awakes only briefly whenever the current audio file changes. The AFS
632 process determines the next audio file, opens it, verifies it has
633 not been changed since it was added to the database and passes the
634 open file descriptor to the server process, along with audio file
635 meta-data such as file name, duration, audio format and so on. The
636 server process then starts to stream the audio file.
638 The AFS process also accepts connections from local clients via
639 a well-known socket. However, only child processes of para_server
640 may connect through this socket. All server commands that have the
641 AFS_READ or AFS_WRITE permission bits use this mechanism to query or
647 *The audio file table*
649 This is the most important and usually also the largest table of the
650 AFS database. It contains the information needed to stream each audio
651 file. In particular the following data is stored for each audio file.
653 - SHA1 hash value of the audio file contents. This is computed
654 once when the file is added to the database. Whenever AFS
655 selects this audio file for streaming the hash value is
656 recomputed and checked against the value stored in the
657 database to detect content changes.
659 - The time when this audio file was last played.
661 - The number of times the file has been played so far.
663 - The attribute bitmask.
665 - The image id which describes the image associated with this
668 - The lyrics id which describes the lyrics associated with
671 - The audio format id (MP3, OGG, ...).
673 - An amplification value that can be used by the amplification
674 filter to pre-amplify the decoded audio stream.
676 - The chunk table. It describes the location and the timing
677 of the building blocks of the audio file. This is used by
678 para_server to send chunks of the file at appropriate times.
680 - The duration of the audio file.
682 - Tag information contained in the audio file (ID3 tags,
683 Vorbis comments, ...).
685 - The number of channels
687 - The encoding bitrate.
689 - The sampling frequency.
691 To add or refresh the data contained in the audio file table, the _add_
692 command is used. It takes the full path of either an audio file or a
693 directory. In the latter case, the directory is traversed recursively
694 and all files which are recognized as valid audio files are added to
697 *The attribute table*
699 The attribute table contains two columns, _name_ and _bitnum_. An
700 attribute is simply a name for a certain bit number in the attribute
701 bitmask of the audio file table.
703 Each of the 64 bits of the attribute bitmask can be set for each
704 audio file individually. Hence up to 64 different attributes may be
705 defined. For example, "pop", "rock", "blues", "jazz", "instrumental",
706 "german_lyrics", "speech", whatever. You are free to choose as
707 many attributes as you like and there are no naming restrictions
710 A new attribute "test" is created by
712 para_client addatt test
716 lists all available attributes. You can set the "test" attribute for
717 an audio file by executing
719 para_client setatt test+ /path/to/the/audio/file
721 Similarly, the "test" bit can be removed from an audio file with
723 para_client setatt test- /path/to/the/audio/file
725 Instead of a path you may use a shell wildcard pattern. The attribute
726 is applied to all audio files matching that pattern:
728 para_client setatt test+ '/test/directory/*'
732 para_client -- ls -lv
734 gives you a verbose listing of your audio files also showing which
737 In case you wonder why the double-dash in the above command is needed:
738 It tells para_client to not interpret the options after the dashes. If
739 you find this annoying, just say
741 alias para='para_client --'
743 and be happy. In what follows we shall use this alias.
745 The "test" attribute can be dropped from the database with
754 for more information and a complete list of command line options to
759 The image, lyrics, moods and playlists tables are all blob tables.
760 Blob tables consist of three columns each: The identifier which is
761 a positive non-negative number that is auto-incremented, the name
762 (an arbitrary string) and the content (the blob).
764 All blob tables support the same set of actions: cat, ls, mv, rm
765 and add. Of course, _add_ is used for adding new blobs to the table
766 while the other actions have the same meaning as the corresponding
767 Unix commands. The paraslash commands to perform these actions are
768 constructed as the concatenation of the table name and the action. For
769 example addimg, catimg, lsimg, mvimg, rmimg are the commands that
770 manipulate or query the image table.
772 The add variant of these commands is special as these commands read
773 the blob contents from stdin. To add an image to the image table the
776 para addimg image_name < file.jpg
780 Note that the images and lyrics are not interpreted at all, and also
781 the playlist and the mood blobs are only investigated when the mood
782 or playlist is activated by using the select command.
786 Unlike all other tables the contents of the score table remain in
787 memory and are never stored on disk. The score table contains two
788 columns: The SHA1 hash value (of an audio file) and its current
791 However, only those files which are admissible for the current mood
792 or playlist are contained in the score table. The audio file selector
793 always chooses the row with the highest score as the file to stream
794 next. While doing so, it computes the new score and updates the
795 last_played and the num_played fields in the audio file table.
797 The score table is recomputed by the select command which loads a
798 new mood or playlist.
803 Playlists and moods offer two different ways of specifying the set of
804 admissible files. A playlist in itself describes a set of admissible
805 files. A mood, in contrast, describes the set of admissible files in
806 terms of attributes and other type of information available in the
807 audio file table. As an example, a mood can define a filename pattern,
808 which is then matched against the names of audio files in the table.
810 Selecting a mood or playlist means the generation of a ranking
811 (a score table) for the set of admissible files. Audio files are
812 then selected on a highest-score-first basis. The score table is
813 recomputed at the moment the mood or playlist is selected.
817 Playlists are accommodated in the playlist table of the afs database,
818 using the aforementioned blob format for tables. A new filelist is
819 created using the addpl command, by specifying the full (absolute)
820 paths of all desired audio files, separated by newlines. For example
822 find /my/mp3/dir -name "*.mp3" | para addpl my_playlist
824 If _my_playlist_ already exists it is overwritten. To activate the
825 new playlist, execute
827 para select p/my_playlist
829 The audio file selector will assign scores to each entry of the list,
830 in descending order so that files will be selected in order. If a
831 file could not be opened for streaming, its entry is removed from
832 the score table (but not from the playlist).
836 A mood consists of a unique name and its *mood definition*, which is
837 a set of *mood lines* containing expressions in terms of attributes
838 and other data contained in the database.
840 At any time, at most one mood can be *active* which means that
841 para_server is going to select only files from that subset of
844 So in order to create a mood definition one has to write a set of
845 mood lines. Mood lines come in three flavours: Accept lines, deny
846 lines and score lines.
848 The general syntax of the three types of mood lines is
851 accept [with score <score>] [if] [not] <mood_method> [options]
852 deny [with score <score>] [if] [not] <mood_method> [options]
853 score <score> [if] [not] <mood_method> [options]
856 Here <score> is either an integer or the string "random" which assigns
857 a random score to all matching files. The score value changes the
858 order in which admissible files are going to be selected, but is of
859 minor importance for this introduction.
861 So we concentrate on the first two forms, i.e. accept and deny
862 lines. As usual, everything in square brackets is optional, i.e.
863 accept/deny lines take the following form when ignoring scores:
865 accept [if] [not] <mood_method> [options]
867 and analogously for the deny case. The "if" keyword is only syntactic
868 sugar and has no function. The "not" keyword just inverts the result,
869 so the essence of a mood line is the mood method part and the options
870 following thereafter.
872 A *mood method* is realized as a function which takes an audio file
873 and computes a number from the data contained in the database.
874 If this number is non-negative, we say the file *matches* the mood
875 method. The file matches the full mood line if it either
877 - matches the mood method and the "not" keyword is not given,
879 - does not match the mood method, but the "not" keyword is given.
881 The set of admissible files for the whole mood is now defined as those
882 files which match at least one accept mood line, but no deny mood line.
883 More formally, an audio file F is admissible if and only if
885 (F ~ AL1 or F ~ AL2...) and not (F ~ DL1 or F ~ DN2 ...)
887 where AL1, AL2... are the accept lines, DL1, DL2... are the deny
888 lines and "~" means "matches".
890 The cases where no mood lines of accept/deny type are defined need
893 - Neither accept nor deny lines: This treats all files as
894 admissible (in fact, that is the definition of the dummy mood
895 which is activated automatically if no moods are available).
897 - Only accept lines: A file is admissible iff it matches at
898 least one accept line:
900 F ~ AL1 or F ~ AL2 or ...
902 - Only deny lines: A file is admissible iff it matches no
905 not (F ~ DL1 or F ~ DN2 ...)
909 *List of mood_methods*
913 Takes no arguments and matches an audio file if and only if no
916 is_set <attribute_name>
918 Takes the name of an attribute and matches iff that attribute is set.
920 path_matches <pattern>
922 Takes a filename pattern and matches iff the path of the audio file
925 artist_matches <pattern>
926 album_matches <pattern>
927 title_matches <pattern>
928 comment_matches <pattern>
930 Takes an extended regular expression and matches iff the text of the
931 corresponding tag of the audio file matches the pattern. If the tag
932 is not set, the empty string is matched against the pattern.
940 Takes a comparator ~ of the set {<, =, <=, >, >=, !=} and a number
941 <num>. Matches an audio file iff the condition <val> ~ <num> is
942 satisfied where val is the corresponding value of the audio file
943 (value of the year tag, bitrate in kbit/s, frequency in Hz, channel
946 The year tag is special as its value is undefined if the audio file
947 has no year tag or the content of the year tag is not a number. Such
948 audio files never match. Another difference is the special treatment
949 if the year tag is a two-digit number. In this case either 1900 or
950 2000 are added to the tag value depending on whether the number is
951 greater than 2000 plus the current year.
956 To create a new mood called "my_mood", write its definition into
957 some temporary file, say "tmpfile", and add it to the mood table
960 para addmood my_mood < tmpfile
962 If the mood definition is really short, you may just pipe it to the
963 client instead of using temporary files. Like this:
965 echo "$MOOD_DEFINITION" | para addmood my_mood
967 There is no need to keep the temporary file since you can always use
968 the catmood command to get it back:
972 A mood can be activated by executing
974 para select m/my_mood
976 Once active, the list of admissible files is shown by the ls command
977 if the "-a" switch is given:
982 *Example mood definition*
984 Suppose you have defined attributes "punk" and "rock" and want to define
985 a mood containing only Punk-Rock songs. That is, an audio file should be
986 admissible if and only if both attributes are set. Since
990 is obviously the same as
992 not (not punk or not rock)
994 (de Morgan's rule), a mood definition that selects only Punk-Rock
997 deny if not is_set punk
998 deny if not is_set rock
1002 File renames and content changes
1003 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1005 Since the audio file selector knows the SHA1 of each audio file that
1006 has been added to the afs database, it recognizes if the content of
1007 a file has changed, e.g. because an ID3 tag was added or modified.
1008 Also, if a file has been renamed or moved to a different location,
1009 afs will detect that an entry with the same hash value already exists
1010 in the audio file table.
1012 In both cases it is enough to just re-add the new file. In the
1013 first case (file content changed), the audio table is updated, while
1014 metadata such as the num_played and last_played fields, as well as
1015 the attributes, remain unchanged. In the other case, when the file
1016 is moved or renamed, only the path information is updated, all other
1017 data remains as before.
1019 It is possible to change the behaviour of the add command by using the
1020 "-l" (lazy add) or the "-f" (force add) option.
1025 Use the debug loglevel (option -l debug for most commands) to show
1026 debugging info. Almost all paraslash executables have a brief online
1027 help which is displayed by using the -h switch. The --detailed-help
1028 option prints the full help text.
1030 If para_server crashed or was killed by SIGKILL (signal 9), it
1031 may refuse to start again because of "dirty osl tables". In this
1032 case you'll have to run the oslfsck program of libosl to fix your
1033 database. It might be necessary to use --force (even if your name
1034 isn't Luke). However, make sure para_server isn't running before
1035 executing oslfsck --force.
1037 If you don't mind to recreate your database you can start
1038 from scratch by removing the entire database directory, i.e.
1040 rm -rf ~/.paraslash/afs_database-0.4
1042 Be aware that this removes all attribute definitions, all playlists
1043 and all mood definitions and requires to re-initialize the tables.
1045 Although oslfsck fixes inconsistencies in database tables it doesn't
1046 care about the table contents. To check for invalid table contents, use
1050 This prints out references to missing audio files as well as invalid
1051 playlists and mood definitions.
1053 ---------------------------------------
1054 Audio formats and audio format handlers
1055 ---------------------------------------
1060 The following audio formats are supported by paraslash:
1064 Mp3, MPEG-1 Audio Layer 3, is a common audio format for audio storage,
1065 designed as part of its MPEG-1 standard. An MP3 file is made up of
1066 multiple MP3 frames, which consist of a header and a data block. The
1067 size of an MP3 frame depends on the bit rate and on the number
1068 of channels. For a typical CD-audio file (sample rate of 44.1 kHz
1069 stereo), encoded with a bit rate of 128 kbit, an MP3 frame is about
1074 OGG is a standardized audio container format, while Vorbis is an
1075 open source codec for lossy audio compression. Since Vorbis is most
1076 commonly made available via the OGG container format, it is often
1077 referred to as OGG/Vorbis. The OGG container format divides data into
1078 chunks called OGG pages. A typical OGG page is about 4KB large. The
1079 Vorbis codec creates variable-bitrate (VBR) data, where the bitrate
1080 may vary considerably.
1084 Speex is an open-source speech codec that is based on CELP (Code
1085 Excited Linear Prediction) coding. It is designed for voice
1086 over IP applications, has modest complexity and a small memory
1087 footprint. Wideband and narrowband (telephone quality) speech are
1088 supported. As for Vorbis audio, Speex bit-streams are often stored
1093 Advanced Audio Coding (AAC) is a standardized, lossy compression
1094 and encoding scheme for digital audio which is the default audio
1095 format for Apple's iPhone, iPod, iTunes. Usually MPEG-4 is used as
1096 the container format and audio files encoded with AAC have the .m4a
1097 extension. A typical AAC frame is about 700 bytes large.
1101 Windows Media Audio (WMA) is an audio data compression technology
1102 developed by Microsoft. A WMA file is usually encapsulated in the
1103 Advanced Systems Format (ASF) container format, which also specifies
1104 how meta data about the file is to be encoded. The bit stream of WMA
1105 is composed of superframes, each containing one or more frames of
1106 2048 samples. For 16 bit stereo a WMA superframe is about 8K large.
1111 Unfortunately, each audio format has its own conventions how meta
1112 data is added as tags to the audio file.
1114 For MP3 files, ID3, version 1 and 2 are widely used. ID3 version 1
1115 is rather simple but also very limited as it supports only artist,
1116 title, album, year and comment tags. Each of these can only be at most
1117 32 characters long. ID3, version 2 is much more flexible but requires
1118 a separate library being installed for paraslash to support it.
1120 Ogg vorbis files contain meta data as Vorbis comments, which are
1121 typically implemented as strings of the form "[TAG]=[VALUE]". Unlike
1122 ID3 version 1 tags, one may use whichever tags are appropriate for
1125 AAC files usually use the MPEG-4 container format for storing meta
1126 data while WMA files wrap meta data as special objects within the
1127 ASF container format.
1129 paraslash only tracks the most common tags that are supported by
1130 all tag variants: artist, title, year, album, comment. When a file
1131 is added to the AFS database, the meta data of the file is extracted
1132 and stored in the audio file table.
1134 Chunks and chunk tables
1135 ~~~~~~~~~~~~~~~~~~~~~~~
1137 paraslash uses the word "chunk" as common term for the building blocks
1138 of an audio file. For MP3 files, a chunk is the same as an MP3 frame,
1139 while for OGG files a chunk is an OGG page, etc. Therefore the chunk
1140 size varies considerably between audio formats, from a few hundred
1141 bytes (MP3) up to 8K (WMA).
1143 The chunk table contains the offsets within the audio file that
1144 correspond to the chunk boundaries of the file. Like the meta data,
1145 the chunk table is computed and stored in the database whenever an
1146 audio file is added.
1148 The paraslash senders (see below) always send complete chunks. The
1149 granularity for seeking is therefore determined by the chunk size.
1151 Audio format handlers
1152 ~~~~~~~~~~~~~~~~~~~~~
1154 For each audio format paraslash contains an audio format handler whose
1155 first task is to tell whether a given file is a valid audio file of
1156 this type. If so, the audio file handler extracts some technical data
1157 (duration, sampling rate, number of channels etc.), computes the
1158 chunk table and reads the meta data.
1160 The audio format handler code is linked into para_server and executed
1161 via the _add_ command. The same code is also available as a stand-alone
1162 tool, para_afh, which can be used to print the technical data, the
1163 chunk table and the meta data of a file. Furthermore, one can use
1164 para_afh to cut an audio file, i.e. to select some of its chunks to
1165 produce a new file containing only these chunks.
1171 Paraslash uses different network connections for control and data.
1172 para_client communicates with para_server over a dedicated TCP control
1173 connection. To transport audio data, separate data connections are
1174 used. For these data connections, a variety of transports (UDP, DCCP,
1175 HTTP) can be chosen.
1177 The chapter starts with the REFERENCE(The paraslash control
1178 service, control service), followed by a section on the various
1179 REFERENCE(Streaming protocols, streaming protocols) in which the data
1180 connections are described. The way audio file headers are embedded into
1181 the stream is discussed REFERENCE(Streams with headers and headerless
1182 streams, briefly) before the REFERENCE(Networking examples, example
1183 section) which illustrates typical commands for real-life scenarios.
1185 Both IPv4 and IPv6 are supported.
1187 The paraslash control service
1188 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1190 para_server is controlled at runtime via the paraslash control
1191 connection. This connection is used for server commands (play, stop,
1192 ...) as well as for afs commands (ls, select, ...).
1194 The server listens on a TCP port and accepts connections from clients
1195 that connect the open port. Each connection causes the server to fork
1196 off a client process which inherits the connection and deals with that
1197 client only. In this classical accept/fork approach the server process
1198 is unaffected if the child dies or goes crazy for whatever reason. In
1199 fact, the child process can not change address space of server process.
1201 The section on REFERENCE(Client-server authentication, client-server
1202 authentication) above described the early connection establishment
1203 from the crypto point of view. Here it is described what happens
1204 after the connection (including crypto setup) has been established.
1205 There are four processes involved during command dispatch as sketched
1206 in the following diagram.
1210 server_host client_host
1211 ~~~~~~~~~~~ ~~~~~~~~~~~
1213 +-----------+ connect +-----------+
1214 |para_server|<------------------------------ |para_client|
1215 +-----------+ +-----------+
1218 +----------> |AFS| |
1222 | | connect (cookie) |
1225 | fork +-----+ inherited connection |
1226 +---------->|child|<--------------------------+
1231 Note that the child process is not a child of the afs process,
1232 so communication of these two processes has to happen via local
1233 sockets. In order to avoid abuse of the local socket by unrelated
1234 processes, a magic cookie is created once at server startup time just
1235 before the server process forks off the AFS process. This cookie is
1236 known to the server, AFS and the child, but not to unrelated processes.
1238 There are two different kinds of commands: First there are commands
1239 that cause the server to respond with some answer such as the list
1240 of all audio files. All but the addblob commands (addimg, addlyr,
1241 addpl, addmood) are of this kind. The addblob commands add contents
1242 to the database, so they need to transfer data the other way round,
1243 from the client to the server.
1245 There is no knowledge about the server commands built into para_client,
1246 so it does not know about addblob commands. Instead, it inspects the
1247 first data package sent by the server for a magic string. If this
1248 string was found, it sends STDIN to the server, otherwise it dumps
1249 data from the server to STDOUT.
1254 A network (audio) stream usually consists of one streaming source,
1255 the _sender_, and one or more _receivers_ which read data over the
1256 network from the streaming source.
1258 Senders are thus part of para_server while receivers are part of
1259 para_audiod. Moreover, there is the stand-alone tool para_recv which
1260 can be used to manually download a stream, either from para_server
1261 or from a web-based audio streaming service.
1263 The following three streaming protocols are supported by paraslash:
1265 - HTTP. Recommended for public streams that can be played by
1266 any player like mpg123, xmms, itunes, winamp, etc. The HTTP
1267 sender is supported on all operating systems and all platforms.
1269 - DCCP. Recommended for LAN streaming. DCCP is currently
1270 available only for Linux.
1272 - UDP. Recommended for multicast LAN streaming.
1274 See the Appendix on REFERENCE(Network protocols, network protocols)
1275 for brief descriptions of the various protocols relevant for network
1276 audio streaming with paraslash.
1278 It is possible to activate more than one sender simultaneously.
1279 Senders can be controlled at run time and via config file and command
1282 Note that audio connections are _not_ encrypted. Transport or Internet
1283 layer encryption should be used if encrypted data connections are
1286 Since DCCP and TCP are both connection-oriented protocols, connection
1287 establishment/teardown and access control are very similar between
1288 these two streaming protocols. UDP is the most lightweight option,
1289 since in contrast to TCP/DCCP it is connectionless. It is also the
1290 only protocol supporting IP multicast.
1292 The HTTP and the DCCP sender listen on a (TCP/DCCP) port waiting for
1293 clients to connect and establish a connection via some protocol-defined
1294 handshake mechanism. Both senders maintain two linked lists each:
1295 The list of all clients which are currently connected, and the list
1296 of access control entries which determines who is allowed to connect.
1297 IP-based access control may be configured through config file and
1298 command line options and via the "allow" and "deny" sender subcommands.
1300 Upon receiving a GET request from the client, the HTTP sender sends
1301 back a status line and a message. The body of this message is the
1302 audio stream. This is common practice and is supported by many popular
1303 clients which can thus be used to play a stream offered by para_server.
1304 For DCCP things are a bit simpler: No messages are exchanged between
1305 the receiver and sender. The client simply connects and the sender
1308 DCCP is an experimental protocol which offers a number of new features
1309 not available for TCP. Both ends can negotiate these features using
1310 a built-in negotiation mechanism. In contrast to TCP/HTTP, DCCP is
1311 datagram-based (no retransmissions) and thus should not be used over
1312 lossy media (e.g. WiFi networks). One useful feature offered by DCCP
1313 is access to a variety of different congestion-control mechanisms
1314 called CCIDs. Two different CCIDs are available per default on Linux:
1317 - _CCID 2_. A Congestion Control mechanism similar to that
1318 of TCP. The sender maintains a congestion window and halves
1319 this window in response to congestion.
1322 - _CCID-3_. Designed to be fair when competing for bandwidth.
1323 It has lower variation of throughput over time compared with
1324 TCP, which makes it suitable for streaming media.
1326 Unlike the HTTP and DCCP senders, the UDP sender maintains only a
1327 single list, the _target list_. This list describes the set of clients
1328 to which the stream is sent. There is no list for access control and
1329 no "allow" and "deny" commands for the UDP sender. Instead, the "add"
1330 and "delete" commands can be used to modify the target list.
1332 Since both UDP and DCCP offer an unreliable datagram-based transport,
1333 additional measures are necessary to guard against disruptions over
1334 networks that are lossy or which may be subject to interference (as
1335 is for instance the case with WiFi). Paraslash uses FEC (Forward
1336 Error Correction) to guard against packet losses and reordering. The
1337 stream is FEC-encoded before it is sent through the UDP socket and
1338 must be decoded accordingly on the receiver side.
1340 The packet size and the amount of redundancy introduced by FEC can
1341 be configured via the FEC parameters which are dictated by server
1342 and may also be configured through the "sender" command. The FEC
1343 parameters are encoded in the header of each network packet, so no
1344 configuration is necessary on the receiver side. See the section on
1345 REFERENCE(Forward error correction, FEC) below.
1347 Streams with headers and headerless streams
1348 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1350 For OGG/Vorbis, OGG/Speex and wma streams, some of the information
1351 needed to decode the stream is only contained in the audio file
1352 header of the container format but not in each data chunk. Clients
1353 must be able to obtain this information in case streaming starts in
1354 the middle of the file or if para_audiod is started while para_server
1355 is already sending a stream.
1357 This is accomplished in different ways, depending on the streaming
1358 protocol. For connection-oriented streams (HTTP, DCCP) the audio file
1359 header is sent prior to audio file data. This technique however does
1360 not work for the connectionless UDP transport. Hence the audio file
1361 header is periodically being embedded into the UDP audio data stream.
1362 By default, the header is resent after five seconds. The receiver has
1363 to wait until the next header arrives before it can start decoding
1369 The sender command of para_server allows to (de-)activate senders
1370 and to change the access permissions senders at runtime. The "si"
1371 (server info) command is used to list the streaming options of the
1372 currently running server as well as the various sender access lists.
1374 -> Show client/target/access lists:
1378 -> Obtain general help for the sender command:
1380 para_client help sender
1382 -> Get help for a specific sender (contains further examples):
1384 s=http # or dccp or udp
1385 para_client sender $s help
1387 By default para_server activates both the HTTP and th DCCP sender on
1388 startup. This can be changed via command line options or para_server's
1391 -> List config file options for senders:
1395 All senders share the "on" and "off" commands, so senders may be
1396 activated and deactivated independently of each other.
1398 -> Switch off the http sender:
1400 para_client sender http off
1402 -> Receive a DCCP stream using CCID2 and write the output into a file:
1404 host=foo.org; ccid=2; filename=bar
1405 para_recv --receiver "dccp --host $host --ccid $ccid" > $filename
1407 Note the quotes around the arguments for the dccp receiver. Each
1408 receiver has its own set of command line options and its own command
1409 line parser, so arguments for the dccp receiver must be protected
1410 from being interpreted by para_recv.
1412 -> Start UDP multicast, using the default multicast address:
1414 para_client sender udp add 224.0.1.38
1416 -> Receive FEC-encoded multicast stream and write the output into a file:
1419 para_recv -r udp > $filename
1421 -> Add an UDP unicast for a client to the target list of the UDP sender:
1424 para_client sender udp add $t
1426 -> Receive this (FEC-encoded) unicast stream:
1429 para_recv -r 'udp -i 0.0.0.0' > $filename
1431 -> Create a minimal config for para_audiod for HTTP streams:
1433 c=$HOME/.paraslash/audiod.conf.min; s=server.foo.com
1434 echo receiver \".:http -i $s\" > $c
1435 para_audiod --config $c
1441 A paraslash filter is a module which transforms an input stream into
1442 an output stream. Filters are included in the para_audiod executable
1443 and in the stand-alone tool para_filter which usually contains the
1446 While para_filter reads its input stream from STDIN and writes
1447 the output to STDOUT, the filter modules of para_audiod are always
1448 connected to a receiver which produces the input stream and a writer
1449 which absorbs the output stream.
1451 Some filters depend on a specific library being installed and are
1452 not compiled in if this library was not found at compile time. To
1453 see the list of supported filters, run para_filter and para_audiod
1454 with the --help option. The output looks similar to the following:
1457 compress wav amp fecdec wmadec prebuffer oggdec aacdec mp3dec
1459 Out of these filter modules, a chain of filters can be constructed,
1460 much in the way Unix pipes can be chained, and analogous to the use
1461 of modules in gstreamer: The output of the first filter becomes the
1462 input of the second filter. There is no limitation on the number of
1463 filters and the same filter may occur more than once.
1465 Like receivers, each filter has its own command line options which
1466 must be quoted to protect them from the command line options of
1467 the driving application (para_audiod or para_filter). Example:
1469 para_filter -f 'mp3dec --ignore-crc' -f 'compress --damp 1'
1471 For para_audiod, each audio format has its own set of filters. The
1472 name of the audio format for which the filter should be applied can
1473 be used as the prefix for the filter option. Example:
1475 para_audiod -f 'mp3:prebuffer --duration 300'
1477 The "mp3" prefix above is actually interpreted as a POSIX extended
1478 regular expression. Therefore
1480 para_audiod -f '.:prebuffer --duration 300'
1482 activates the prebuffer filter for all supported audio formats (because
1483 "." matches all audio formats) while
1485 para_audiod -f 'wma|ogg:prebuffer --duration 300'
1487 activates it only for wma and ogg streams.
1492 For each supported audio format there is a corresponding filter
1493 which decodes audio data in this format to 16 bit PCM data which
1494 can be directly sent to the sound device or any other software that
1495 operates on undecoded PCM data (visualizers, equalizers etc.). Such
1496 filters are called _decoders_ in general, and xxxdec is the name of
1497 the paraslash decoder for the audio format xxx. For example, the mp3
1498 decoder filter is called mp3dec.
1500 Note that the output of the decoder is about 10 times larger than
1501 its input. This means that filters that operate on the decoded audio
1502 stream have to deal with much more data than filters that transform
1503 the audio stream before it is fed to the decoder.
1505 Paraslash relies on external libraries for most decoders, so these
1506 libraries must be installed for the decoder to be included in the
1507 para_filter and para_audiod executables. The oggdec filter depends
1508 on the libogg and libvorbis libraries for example.
1510 Forward error correction
1511 ~~~~~~~~~~~~~~~~~~~~~~~~
1513 As already mentioned REFERENCE(Streaming protocols, earlier),
1514 paraslash uses forward error correction (FEC) for the unreliable UDP
1515 and DCCP transports. FEC is a technique which was invented already
1516 in 1960 by Reed and Solomon and which is widely used for the parity
1517 calculations of storage devices (RAID arrays). It is based on the
1518 algebraic concept of finite fields, today called Galois fields, in
1519 honour of the mathematician Galois (1811-1832). The FEC implementation
1520 of paraslash is based on code by Luigi Rizzo.
1522 Although the details require a sound knowledge of the underlying
1523 mathematics, the basic idea is not hard to understand: For positive
1524 integers k and n with k < n it is possible to compute for any k given
1525 data bytes d_1, ..., d_k the corresponding r := n -k parity bytes p_1,
1526 ..., p_r such that all data bytes can be reconstructed from *any*
1529 {d_1, ..., d_k, p_1, ..., p_r}.
1531 FEC-encoding for unreliable network transports boils down to slicing
1532 the audio stream into groups of k suitably sized pieces called _slices_
1533 and computing the r corresponding parity slices. This step is performed
1534 in para_server which then sends both the data and the parity slices
1535 over the unreliable network connection. If the client was able
1536 to receive at least k of the n = k + r slices, it can reconstruct
1537 (FEC-decode) the original audio stream.
1539 From these observations it is clear that there are three different
1540 FEC parameters: The slice size, the number of data slices k, and the
1541 total number of slices n. It is crucial to choose the slice size
1542 such that no fragmentation of network packets takes place because
1543 FEC only guards against losses and reordering but fails if slices are
1546 FEC decoding in paralash is performed through the fecdec filter which
1547 usually is the first filter (there can be other filters before fecdec
1548 if these do not alter the audio stream).
1551 Volume adjustment (amp and compress)
1552 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1554 The amp and the compress filter both adjust the volume of the audio
1555 stream. These filters operate on uncompressed audio samples. Hence
1556 they are usually placed directly after the decoding filter. Each
1557 sample is multiplied with a scaling factor (>= 1) which makes amp
1558 and compress quite expensive in terms of computing power.
1562 The amp filter amplifies the audio stream by a fixed scaling factor
1563 that must be known in advance. For para_audiod this factor is derived
1564 from the amplification field of the audio file's entry in the audio
1565 file table while para_filter uses the value given at the command line.
1567 The optimal scaling factor F for an audio file is the largest real
1568 number F >= 1 such that after multiplication with F all samples still
1569 fit into the sample interval [-32768, 32767]. One can use para_filter
1570 in combination with the sox utility to compute F:
1572 para_filter -f mp3dec -f wav < file.mp3 | sox -t wav - -e stat -v
1574 The amplification value V which is stored in the audio file table,
1575 however, is an integer between 0 and 255 which is connected to F
1580 To store V in the audio file table, the command
1582 para_client -- touch -a=V file.mp3
1584 is used. The reader is encouraged to write a script that performs
1585 these computations :)
1589 Unlike the amplification filter, the compress filter adjusts the volume
1590 of the audio stream dynamically without prior knowledge about the peak
1591 value. It maintains the maximal volume of the last n samples of the
1592 audio stream and computes a suitable amplification factor based on that
1593 value and the various configuration options. It tries to chose this
1594 factor such that the adjusted volume meets the desired target level.
1596 Note that it makes sense to combine amp and compress.
1598 Misc filters (wav and prebuffer)
1599 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1601 These filters are rather simple and do not modify the audio stream at
1602 all. The wav filter is only useful with para_filter and in connection
1603 with a decoder. It asks the decoder for the number of channels and the
1604 sample rate of the stream and adds a Microsoft wave header containing
1605 this information at the beginning. This allows to write wav files
1606 rather than raw PCM files (which do not contain any information about
1607 the number of channels and the sample rate).
1609 The prebuffer filter simply delays the output until the given time has
1610 passed (starting from the time the first byte was available in its
1611 input queue) or until the given amount of data has accumulated. It
1612 is mainly useful for para_audiod if the standard parameters result
1613 in buffer underruns.
1615 Both filters require almost no additional computing time, even when
1616 operating on uncompressed audio streams, since data buffers are simply
1617 "pushed down" rather than copied.
1622 -> Decode an mp3 file to wav format:
1624 para_filter -f mp3dec -f wav < file.mp3 > file.wav
1626 -> Amplify a raw audio file by a factor of 1.5:
1628 para_filter -f amp --amp 32 < foo.raw > bar.raw
1634 Once an audio stream has been received and decoded to PCM format,
1635 it can be sent to a sound device for playback. This part is performed
1636 by paraslash _writers_ which are described in this chapter.
1641 A paraslash writer acts as a data sink that consumes but does not
1642 produce audio data. Paraslash writers operate on the client side and
1643 are contained in para_audiod and in the stand-alone tool para_write.
1645 The para_write program reads uncompressed audio data from STDIN. If
1646 this data starts with a wav header, sample rate, sample format and
1647 channel count are read from the header. Otherwise CD audio (44.1KHz
1648 16 bit little endian, stereo) is assumed but this can be overridden
1649 by command line options. para_audiod, on the other hand, obtains
1650 the sample rate and the number of channels from the decoder.
1652 Like receivers and filters, each writer has an individual set of
1653 command line options, and for para_audiod writers can be configured
1654 per audio format separately. It is possible to activate more than
1655 one writer for the same stream simultaneously.
1660 Unfortunately, the various flavours of Unix on which paraslash
1661 runs on have different APIs for opening a sound device and starting
1662 playback. Hence for each such API there is a paraslash writer that
1663 can play the audio stream via this API.
1665 *ALSA*. The _Advanced Linux Sound Architecture_ is only available on
1666 Linux systems. Although there are several mid-layer APIs in use by
1667 the various Linux distributions (ESD, Jack, PulseAudio), paraslash
1668 currently supports only the low-level ALSA API which is not supposed
1669 to be change. ALSA is very feature-rich, in particular it supports
1670 software mixing via its DMIX plugin. ALSA is the default writer on
1673 *OSS*. The _Open Sound System_ is the only API on *BSD Unixes and
1674 is also available on Linux systems, usually provided by ALSA as an
1675 emulation for backwards compatibility. This API is rather simple but
1676 also limited. For example only one application can open the device
1677 at any time. The OSS writer is activated by default on BSD Systems.
1679 *OSX*. Mac OS X has yet another API called CoreAudio. The OSX writer
1680 for this API is only compiled in on such systems and is of course
1683 *FILE*. The file writer allows to capture the audio stream and
1684 write the PCM data to a file on the file system rather than playing
1685 it through a sound device. It is supported on all platforms and is
1688 *AO*. _Libao_ is a cross-platform audio library which supports a wide
1689 variety of platforms including PulseAudio (gnome), ESD (Enlightened
1690 Sound Daemon), AIX, Solaris and IRIX. The ao writer plays audio
1691 through an output plugin of libao.
1696 -> Use the OSS writer to play a wav file:
1698 para_write --writer oss < file.wav
1700 -> Enable ALSA software mixing for mp3 streams
1702 para_audiod --writer 'mp3:alsa -d plug:swmix'
1709 para_gui executes an arbitrary command which is supposed to print
1710 status information to STDOUT. It then displays this information in
1711 a curses window. By default the command
1713 para_audioc -- stat -p
1715 is executed, but this can be customized via the --stat_cmd option. In
1716 particular it possible to use
1718 para_client -- stat -p
1720 to make para_gui work on systems on which para_audiod is not running.
1725 It is possible to bind keys to arbitrary commands via custom
1726 key-bindings. Besides the internal keys which can not be changed (help,
1727 quit, loglevel, version...), the following flavours of key-bindings
1730 - external: Shutdown curses before launching the given command.
1731 Useful for starting other ncurses programs from within
1732 para_gui, e.g. aumix or dialog scripts. Or, use the mbox
1733 output format to write a mailbox containing one mail for each
1734 (admissible) file the audio file selector knows about. Then
1735 start mutt from within para_gui to browse your collection!
1737 - display: Launch the command and display its stdout in
1738 para_gui's bottom window.
1740 - para: Like display, but start "para_client <specified
1741 command>" instead of "<specified command>".
1743 The general form of a key binding is
1747 which maps key k to command c using mode m. Mode may be x, d or p
1748 for external, display and paraslash commands, respectively.
1753 Currently there are only two themes for para_gui. It is easy, however,
1754 to add more themes. To create a new theme one has to define the
1755 position, color and geometry for for each status item that should be
1756 shown by this theme. See gui_theme.c for examples.
1758 The "." and "," keys are used to switch between themes.
1763 -> Show server info:
1767 -> Jump to the middle of the current audio file by pressing F5:
1769 key_map "<F5>:p:jmp 50"
1771 -> vi-like bindings for jumping around:
1774 key_map "h:p:ff 10-"
1776 key_map "b:p:ff 60-"
1778 -> Print the current date and time:
1782 -> Call other curses programs:
1785 key_map "!:x:/bin/bash"
1786 key_map "^E:x:/bin/sh -c 'vi ~/.paraslash/gui.conf'"
1795 In order to compile the sources from the git repository (rather than
1796 from tar balls) and for contributing non-trivial changes to the
1797 paraslash project, some additional tools should be installed on a
1800 http://git.or.cz/ (git). As described in more detail REFERENCE(Git
1801 branches, below), the git source code management tool is used for
1802 paraslash development. It is necessary for cloning the git repository
1803 and for getting updates.
1805 ftp://ftp.gnu.org/pub/gnu/gengetopt/ (gengetopt). The C code for
1806 the command line parsers of all paraslash executables is generated
1807 by gengetopt. The generated C files are shipped in the tarballs but
1808 are not contained in the git repository.
1810 ftp://ftp.gnu.org/pub/gnu/m4/ (m4). Some input files for gengetopt
1811 are generated from templates by the m4 macro processor.
1813 ftp://ftp.gnu.org/pub/gnu/autoconf/ (autoconf) GNU autoconf creates
1814 the configure file which is shipped in the tarballs but has to be
1815 generated when compiling from git.
1817 http://www.triptico.com/software/grutatxt.html (grutatxt). The
1818 HTML version of this manual and some of the paraslash web pages are
1819 generated by the grutatxt plain text to HTML converter. If changes
1820 are made to these text files the grutatxt package must be installed
1821 to regenerate the HTML files.
1823 http://www.stack.nl/~dimitri/doxygen/ (doxygen). The documentation
1824 of paraslash's C sources uses the doxygen documentation system. The
1825 conventions for documenting the source code is described in the
1826 REFERENCE(Doxygen, Doxygen section).
1828 ftp://ftp.gnu.org/pub/gnu/global (global). This is used to generate
1829 browsable HTML from the C sources. It is needed by doxygen.
1834 Paraslash has been developed using the git source code management
1835 tool since 2006. Development is organized roughly in the same spirit
1836 as the git development itself, as described below.
1838 The following text passage is based on "A note from the maintainer",
1839 written by Junio C Hamano, the maintainer of git.
1841 There are four branches in the paraslash repository that track the
1842 source tree: "master", "maint", "next", and "pu".
1844 The "master" branch is meant to contain what is well tested and
1845 ready to be used in a production setting. There could occasionally be
1846 minor breakages or brown paper bag bugs but they are not expected to
1847 be anything major, and more importantly quickly and easily fixable.
1848 Every now and then, a "feature release" is cut from the tip of this
1849 branch, named with three dotted decimal digits, like 0.4.2.
1851 Whenever changes are about to be included that will eventually lead to
1852 a new major release (e.g. 0.5.0), a "maint" branch is forked off from
1853 "master" at that point. Obvious, safe and urgent fixes after the major
1854 release are applied to this branch and maintenance releases are cut
1855 from it. New features never go to this branch. This branch is also
1856 merged into "master" to propagate the fixes forward.
1858 A trivial and safe enhancement goes directly on top of "master".
1859 New development does not usually happen on "master", however.
1860 Instead, a separate topic branch is forked from the tip of "master",
1861 and it first is tested in isolation; Usually there are a handful such
1862 topic branches that are running ahead of "master". The tip of these
1863 branches is not published in the public repository to keep the number
1864 of branches that downstream developers need to worry about low.
1866 The quality of topic branches varies widely. Some of them start out as
1867 "good idea but obviously is broken in some areas" and then with some
1868 more work become "more or less done and can now be tested by wider
1869 audience". Luckily, most of them start out in the latter, better shape.
1871 The "next" branch is to merge and test topic branches in the latter
1872 category. In general, this branch always contains the tip of "master".
1873 It might not be quite rock-solid production ready, but is expected to
1874 work more or less without major breakage. The maintainer usually uses
1875 the "next" version of paraslash for his own pleasure, so it cannot
1876 be _that_ broken. The "next" branch is where new and exciting things
1879 The two branches "master" and "maint" are never rewound, and "next"
1880 usually will not be either (this automatically means the topics that
1881 have been merged into "next" are usually not rebased, and you can find
1882 the tip of topic branches you are interested in from the output of
1883 "git log next"). You should be able to safely build on top of them.
1885 However, at times "next" will be rebuilt from the tip of "master" to
1886 get rid of merge commits that will never be in "master. The commit
1887 that replaces "next" will usually have the identical tree, but it
1888 will have different ancestry from the tip of "master".
1890 The "pu" (proposed updates) branch bundles the remainder of the
1891 topic branches. The "pu" branch, and topic branches that are only in
1892 "pu", are subject to rebasing in general. By the above definition
1893 of how "next" works, you can tell that this branch will contain quite
1894 experimental and obviously broken stuff.
1896 When a topic that was in "pu" proves to be in testable shape, it
1897 graduates to "next". This is done with
1900 git merge that-topic-branch
1902 Sometimes, an idea that looked promising turns out to be not so good
1903 and the topic can be dropped from "pu" in such a case.
1905 A topic that is in "next" is expected to be polished to perfection
1906 before it is merged to "master". Similar to the above, this is
1910 git merge that-topic-branch
1911 git branch -d that-topic-branch
1913 Note that being in "next" is not a guarantee to appear in the next
1914 release (being in "master" is such a guarantee, unless it is later
1915 found seriously broken and reverted), nor even in any future release.
1920 The preferred coding style for paraslash coincides more or less
1921 with the style of the Linux kernel. So rather than repeating what is
1922 written XREFERENCE(http://www.kernel.org/doc/Documentation/CodingStyle,
1923 there), here are the most important points.
1925 - Burn the GNU coding standards.
1926 - Never use spaces for indentation.
1927 - Tabs are 8 characters, and thus indentations are also 8 characters.
1928 - Don't put multiple assignments on a single line.
1929 - Avoid tricky expressions.
1930 - Don't leave whitespace at the end of lines.
1931 - The limit on the length of lines is 80 columns.
1932 - Use K&R style for placing braces and spaces:
1938 - Use a space after (most) keywords.
1939 - Do not add spaces around (inside) parenthesized expressions.
1940 - Use one space around (on each side of) most binary and ternary operators.
1941 - Do not use cute names like ThisVariableIsATemporaryCounter, call it tmp.
1942 - Mixed-case names are frowned upon.
1943 - Descriptive names for global variables are a must.
1945 - Functions should be short and sweet, and do just one thing.
1946 - The number of local variables shouldn't exceed 10.
1947 - Gotos are fine if they improve readability and reduce nesting.
1948 - Don't use C99-style "// ..." comments.
1949 - Names of macros defining constants and labels in enums are capitalized.
1950 - Enums are preferred when defining several related constants.
1951 - Always use the paraslash wrappers for allocating memory.
1952 - If the name of a function is an action or an imperative.
1953 command, the function should return an error-code integer
1954 (<0 means error, >=0 means success). If the name is a
1955 predicate, the function should return a "succeeded" boolean.
1961 Doxygen is a documentation system for various programming
1962 languages. The paraslash project uses Doxygen for generating the API
1963 reference on the web pages, but good source code documentation is
1964 also beneficial to people trying to understand the code structure
1965 and the interactions between the various source files.
1967 It is more illustrative to look at the source code for examples than
1968 to describe the conventions for documenting the source in this manual,
1969 so we only describe which parts of the code need doxygen comments,
1970 but leave out details on documentation conventions.
1972 As a rule, only the public part of the C source is documented with
1973 Doxygen. This includes structures, defines and enumerations in header
1974 files as well as public (non-static) C functions. These should be
1975 documented completely. For example each parameter and the return
1976 value of a public function should get a descriptive comment.
1978 No doxygen comments are necessary for static functions and for
1979 structures and enumerations in C files (which are used only within
1980 this file). This does not mean, however, that those entities need
1981 no documentation at all. Instead, common sense should be applied to
1982 document what is not obvious from reading the code.
1991 *IP*. The _Internet Protocol_ is the primary networking protocol
1992 used for the Internet. All protocols described below use IP as the
1993 underlying layer. Both the prevalent IPv4 and the next-generation
1994 IPv6 variant are being deployed actively worldwide.
1996 *Connection-oriented and connectionless protocols*. Connectionless
1997 protocols differ from connection-oriented ones in that state
1998 associated with the sending/receiving endpoints is treated
1999 implicitly. Connectionless protocols maintain no internal knowledge
2000 about the state of the connection. Hence they are not capable of
2001 reacting to state changes, such as sudden loss or congestion on the
2002 connection medium. Connection-oriented protocols, in contrast, make
2003 this knowledge explicit. The connection is established only after
2004 a bidirectional handshake which requires both endpoints to agree
2005 on the state of the connection, and may also involve negotiating
2006 specific parameters for the particular connection. Maintaining an
2007 up-to-date internal state of the connection also in general means
2008 that the sending endpoints perform congestion control, adapting to
2009 qualitative changes of the connection medium.
2011 *Reliability*. In IP networking, packets can be lost, duplicated,
2012 or delivered out of order, and different network protocols handle
2013 these problems in different ways. We call a transport-layer protocol
2014 _reliable_, if it turns the unreliable IP delivery into an ordered,
2015 duplicate- and loss-free delivery of packets. Sequence numbers
2016 are used to discard duplicates and re-arrange packets delivered
2017 out-of-order. Retransmission is used to guarantee loss-free
2018 delivery. Unreliable protocols, in contrast, do not guarantee ordering
2021 *Classification*. With these definitions the protocols which are used
2022 by paraslash for steaming audio data may be classified as follows.
2024 - HTTP/TCP: connection-oriented, reliable,
2025 - UDP: connectionless, unreliable,
2026 - DCCP: connection-oriented, unreliable.
2028 Below we give a short descriptions of these protocols.
2030 *TCP*. The _Transmission Control Protocol_ provides reliable,
2031 ordered delivery of a stream and a classic window-based congestion
2032 control. In contrast to UDP and DCCP (see below), TCP does not have
2033 record-oriented or datagram-based syntax, i.e. it provides a stream
2034 which is unaware and independent of any record (packet) boundaries.
2035 TCP is used extensively by many application layers. Besides HTTP (the
2036 Hypertext Transfer Protocol), also FTP (the File Transfer protocol),
2037 SMTP (Simple Mail Transfer Protocol), SSH (Secure Shell) all sit on
2040 *UDP*. The _User Datagram Protocol_ is the simplest transport-layer
2041 protocol, built as a thin layer directly on top of IP. For this reason,
2042 it offers the same best-effort service as IP itself, i.e. there is no
2043 detection of duplicate or reordered packets. Being a connectionless
2044 protocol, only minimal internal state about the connection is
2045 maintained, which means that there is no protection against packet
2046 loss or network congestion. Error checking and correction (if at all)
2047 are performed in the application.
2049 *DCCP*. The _Datagram Congestion Control Protocol_ combines the
2050 connection-oriented state maintenance known from TCP with the
2051 unreliable, datagram-based transport of UDP. This means that it
2052 is capable of reacting to changes in the connection by performing
2053 congestion control, offering multiple alternative approaches. But it
2054 is bound to datagram boundaries (the maximum packet size supported
2055 by a medium), and like UDP it lacks retransmission to protect
2056 against loss. Due to the use of sequence numbers, it is however
2057 able to react to loss (interpreted as a congestion indication) and
2058 to ignore out-of-order and duplicate packets. Unlike TCP it allows
2059 to negotiate specific, binding features for a connection, such as
2060 the choice of congestion control: classic, window-based congestion
2061 control known from TCP is available as CCID-2, rate-based, "smooth"
2062 congestion control is offered as CCID-3.
2064 *HTTP*. _The Hypertext Transfer Protocol_ is an application layer
2065 protocol on top of TCP. It is spoken by web servers and is most often
2066 used for web services. However, as can be seen by the many Internet
2067 radio stations and YouTube/Flash videos, http is by far not limited to
2068 the delivery of web pages only. Being a simple request/response based
2069 protocol, the semantics of the protocol also allow the delivery of
2070 multimedia content, such as audio over http.
2072 *Multicast*. IP multicast is not really a protocol but a technique
2073 for one-to-many communication over an IP network. The challenge is to
2074 deliver information to a group of destinations simultaneously using
2075 the most efficient strategy to send the messages over each link of
2076 the network only once. This has benefits for streaming multimedia:
2077 the standard one-to-one unicast offered by TCP/DCCP means that
2078 n clients listening to the same stream also consume n-times the
2079 resources, whereas multicast requires to send the stream just once,
2080 irrespective of the number of receivers. Since it would be costly to
2081 maintain state for each listening receiver, multicast often implies
2082 connectionless transport, which is the reason that it is currently
2083 only available via UDP.
2088 Paraslash is licensed under the GPL, version 2. Most of the code
2089 base has been written from scratch, and those parts are GPL V2
2090 throughout. Notable exceptions are FEC and the WMA decoder. See the
2091 corresponding source files for licencing details for these parts. Some
2092 code sniplets of several other third party software packages have
2093 been incorporated into the paraslash sources, for example log message
2094 coloring was taken from the git sources. These third party software
2095 packages are all published under the GPL or some other license
2096 compatible to the GPL.
2101 Many thanks to Gerrit Renker who read an early draft of this manual
2102 and contributed significant improvements.
2110 - Reed, Irving S.; Solomon, Gustave (1960),
2111 XREFERENCE(http://kom.aau.dk/~heb/kurser/NOTER/KOFA01.PDF,
2112 Polynomial Codes over Certain Finite Fields), Journal of the
2113 Society for Industrial and Applied Mathematics (SIAM) 8 (2):
2114 300-304, doi:10.1137/0108018)
2119 - XREFERENCE(http://www.ietf.org/rfc/rfc768.txt, RFC 768) (1980):
2120 User Datagram Protocol
2121 - XREFERENCE(http://www.ietf.org/rfc/rfc791.txt, RFC 791) (1981):
2123 - XREFERENCE(http://www.ietf.org/rfc/rfc2437.txt, RFC 2437) (1998):
2124 RSA Cryptography Specifications
2125 - XREFERENCE(http://www.ietf.org/rfc/rfc4340.txt, RFC 4340)
2126 (2006): Datagram Congestion Control Protocol (DCCP)
2127 - XREFERENCE(http://www.ietf.org/rfc/rfc4341.txt, RFC 4341) (2006):
2128 Congestion Control ID 2: TCP-like Congestion Control
2129 - XREFERENCE(http://www.ietf.org/rfc/rfc4342.txt, RFC 4342) (2006):
2130 Congestion Control ID 3: TCP-Friendly Rate Control (TFRC)
2132 Application web pages
2133 ~~~~~~~~~~~~~~~~~~~~~
2135 - XREFERENCE(http://paraslash.systemlinux.org/, paraslash)
2136 - XREFERENCE(http://xmms2.org/wiki/Main_Page, xmms)
2137 - XREFERENCE(http://www.mpg123.de/, mpg123)
2138 - XREFERENCE(http://gstreamer.freedesktop.org/, gstreamer)
2139 - XREFERENCE(http://www.icecast.org/, icecast)
2140 - XREFERENCE(http://beesbuzz.biz/code/audiocompress.php, Audio Compress)
2142 External documentation
2143 ~~~~~~~~~~~~~~~~~~~~~~
2145 - XREFERENCE(http://kernel.org/pub/linux/kernel/people/hpa/raid6.pdf,
2146 H. Peter Anvin: The mathematics of Raid6)
2147 - XREFERENCE(http://info.iet.unipi.it/~luigi/fec_ccr.ps.gz,
2148 Luigi Rizzo: Effective Erasure Codes for reliable Computer
2149 Communication Protocols)
2153 - XREFERENCE(http://info.iet.unipi.it/~luigi/vdm.tar.gz,
2154 Original FEC implementation by Luigi Rizzo)