cdda2wav - a sampling utility that dumps CD audio data into wav sound
cdda2wav [-c chans] [-s] [-m] [-b bits] [-r rate] [-a divider] [-t
track[+endtrack]] [-i index] [-o offset] [-d duration] [-x] [-q] [-w]
[-v optlist] [-V] [-Q] [-J] [-L cddbmode] [-R] [-P sectors] [-F] [-G]
[-T] [-e] [-p percentage] [-n sectors] [-l buffers] [-N] [-J] [-H] [-g]
[-B] [-D device] [-A auxdevice] [-I interface] [-O audiotype] [-C
input-endianess] [-E output-endianess] [-M count] [-S speed] [-para-
noia] [cddbp-server=servername] [cddbp-port=portnumber] [filename(s) or
cdda2wav can retrieve audio tracks from CDROM drives (see README for a
list of drives) that are capable of reading audio data digitally to the
uses device as the source for CDDA reading. For example
/dev/cdrom for the cooked_ioctl interface and Bus,ID,Lun for the
generic_scsi interface. The device has to correspond with the
interface setting (see below).
Using the cooked_ioctl is not recommended as this makes cdda2wav
mainly depend on the audio extraction quality of the operating
system which is usually extremely bad.
The setting of the environment variable CDDA_DEVICE is overrid-
den by this option.
uses auxdevice as CDROM drive for ioctl usage.
specifies the interface for CDROM access: generic_scsi or (on
Linux, and FreeBSD systems) cooked_ioctl.
-c channels --channels
uses 1 for mono, or 2 for stereo recording, or s for stereo
recording with both channels swapped.
sets to stereo recording.
sets to mono recording.
sets maximum (CD) quality.
-b bits --bits-per-sample
sets bits per sample per channel: 8, 12 or 16.
-r rate --rate
sets rate in samples per second. Possible values are listed
with the -R option.
-a divider --divider
sets rate to 44100Hz / divider. Possible values are listed with
the -R option.
shows a list of all sample rates and their dividers.
-P sectors --set-overlap
sets the initial number of overlap sectors for jitter correc-
-n sectors --sectors-per-request
reads sectors per request.
-l buffers --buffers-in-ring
uses a ring buffer with buffers total.
-t track+endtrack --track
selects the start track and optionally the end track.
-i index --index
selects the start index.
-o offset --offset
starts offset sectors behind start track (one sector equivalents
-O audiotype --output-format
can be wav (for wav files) or aiff (for apple/sgi aiff files) or
aifc (for apple/sgi aifc files) or au or sun (for sun .au PCM
files) or cdr or raw (for headerless files to be used for cd
-C endianess --cdrom-endianess
sets endianess of the input samples to ’little’, ’big’ or
’guess’ to override defaults.
-E endianess --output-endianess
sets endianess of the output samples to ’little’ or ’big’ to
-d duration --duration
sets recording time in seconds or frames. Frames (sectors) are
indicated by a ’f’ suffix (like 75f for 75 sectors). 0 sets the
time for whole track.
-B --bulk --alltracks
copies each track into a seperate file.
waits for signal, then start recording.
finds extrem amplitudes in samples.
finds if input samples are in mono.
undo the effect of pre-emphasis in the input samples.
copies audio data to sound device e.g. /dev/dsp.
-p percentage --set-pitch
changes pitch of audio data copied to sound device.
-v itemlist --verbose-level
prints verbose information about the CD. Level is a list of
comma seperated suboptions. Each suboption controls the type of
information to be reported.
|Suboption | Description |
| disable | no information is given, warnings appear however |
| all | all information is given |
| toc | show table of contents |
| summary | show a summary of the recording parameters |
| indices | determine and display index offsets |
| catalog | retrieve and display the media catalog number MCN |
| trackid | retrieve and display all Intern. Standard Recording Codes ISRC |
| sectors | show the table of contents in start sector notation |
| titles | show the table of contents with track titles (when available) |
does not write to a file, it just reads (for debugging pur-
does not write to a file, it just gives information about the
-L cddb mode --cddb
does a cddbp album- and track title lookup based on the cddb id.
The parameter cddb mode defines how multiple entries shall be
|Parameter | Description |
| 0 | interactive mode. The user selects the entry to use. |
| 1 | first fit mode. The first entry is taken unconditionally. |
sets the server to be contacted for title lookups.
sets the port number to be used for title lookups.
does not write an info file and a cddb file.
formats the output to be better parsable by gui frontends.
-M count --md5
enables calculation of MD-5 checksum for ’count’ bytes from a
beginning of a track.
-S speed --speed
sets the cdrom device to one of the selectable speeds for read-
quiet operation, no screen output.
enable SCSI command logging to the console. This is mainly used
suppress SCSI command error reports to the console. This is
mainly used for guis.
Scan all SCSI devices on all SCSI busses and print the inquiry
strings. This option may be used to find SCSI address of the
CD/DVD-Recorder on a system. The numbers printed out as labels
are computed by: bus * 100 + target
use the paranoia library instead of cdda2wav’s routines for
display version of cdda2wav on standard output.
Defaults depend on the
Makefile and environment variable settings (currently
CDDA_DEVICE is used to set the device name. The device naming is com-
patible with Joerg Schilling’s cdrecord package.
is used for cddbp title lookups when supplied.
is used for cddbp title lookups when supplied.
RSH If the RSH environment variable is present, the remote connec-
tion will not be created via rcmd(3) but by calling the program
pointed to by RSH. Use e.g. RSH=/usr/bin/ssh to create a
secure shell connection.
Note that this forces cdda2wav to create a pipe to the rsh(1)
program and disallows cdda2wav to directly access the network
socket to the remote server. This makes it impossible to set up
performance parameters and slows down the connection compared to
a root initiated rcmd(3) connection.
RSCSI If the RSCSI environment variable is present, the remote SCSI
server will not be the program /opt/schily/sbin/rscsi but the
program pointed to by RSCSI. Note that the remote SCSI server
program name will be ignored if you log in using an account that
has been created with a remote SCSI server program as login
cdda2wav uses the following exit codes to indicate various degress of
|Exitcode | Description |
| 0 | no errors encountered, successful operation. |
| 1 | usage or syntax error. cdda2wav got inconsistent arguments. |
| 2 | permission (un)set errors. permission changes failed. |
| 3 | read errors on the cdrom/burner device encountered. |
| 4 | write errors while writing one of the output files encountered. |
| 5 | errors with soundcard handling (initialization/write). |
| 6 | errors with stat() system call on the read device (cooked ioctl). |
| 7 | pipe communication errors encountered (in forked mode). |
| 8 | signal handler installation errors encountered. |
| 9 | allocation of shared memory failed (in forked mode). |
| 10 | dynamic heap memory allocation failed. |
| 11 | errors on the audio cd medium encountered. |
| 12 | device open error in ioctl handling detected. |
| 13 | race condition in ioctl interface handling detected. |
| 14 | error in ioctl() operation encountered. |
| 15 | internal error encountered. Please report back!!! |
| 16 | error in semaphore operation encountered (install / request). |
| 17 | could not get the scsi transfer buffer. |
| 18 | could not create pipes for process communication (in forked mode). |
cdda2wav is able to read parts of an audio CD or multimedia CDROM (con-
taining audio parts) directly digitally. These parts can be written to
a file, a pipe, or to a sound device.
cdda2wav stands for CDDA to WAV (where CDDA stands for compact disc
digital audio and WAV is a sound sample format introduced by MS Win-
dows). It allows copying CDDA audio data from the CDROM drive into a
file in WAV or other formats.
The latest versions try to get higher real-time scheduling priorities
to ensure smooth (uninterrupted) operation. These priorities are avail-
able for super users and are higher than those of ’normal’ processes.
Thus delays are minimized.
If your CDROM is on device DEV and it is loaded with an audio CD, you
may simply invoke cdda2wav dev=DEV and it will create the sound file
audio.wav recording the whole track beginning with track 1 in stereo at
16 bit at 44100 Hz sample rate, if your file system has enough space
free. Otherwise recording time will be limited. For details see files
README and README.INSTALL
HINTS ON OPTIONS
Most of the options are used to control the format of the WAV
file. In the following text all of them are described.
-D device selects the CDROM drive device to be used. The speci-
fier given should correspond to the selected interface (see
below). CHANGE! For the cooked_ioctl interface this is the
cdrom device descriptor as before. The SCSI devices used with
the generic SCSI interface however are now addressed with their
SCSI-Bus, SCSI-Id, and SCSI-Lun instead of the generic SCSI
device descriptor!!! One example for a SCSI CDROM drive on bus
0 with SCSI ID 3 and lun 0 is -D0,3,0.
Select Auxiliary device
-A auxdevice is necessary for CD-Extra handling. For Non-SCSI-
CDROM drives this is the same device as given by -D (see above).
For SCSI-CDROM drives it is the CDROM drive (SCSI) device (i.e.
/dev/sr0 ) corresponding to the SCSI device (i.e. 0,3,0 ). It
has to match the device used for sampling.
-I interface selects the CDROM drive interface. For SCSI drives
use generic_scsi (cooked_ioctl may not yet be available for all
devices): generic_scsi and cooked_ioctl. The first uses the
generic SCSI interface, the latter uses the ioctl of the CDROM
driver. The latter variant works only when the kernel driver
supports CDDA reading. This entry has to match the selected
CDROM device (see above).
Enable echo to soundcard
-e copies audio data to the sound card while recording, so you
hear it nearly simultaneously. The soundcard gets the same data
that is recorded. This is time critical, so it works best with
the -q option. To use cdda2wav as a pseudo CD player without
recording in a file you could use cdda2wav -q -e -t2 -d0 -N to
play the whole second track. This feature reduces the recording
speed to at most onefold speed. You cannot make better record-
ings than your sound card can play (since the same data is
Change pitch of echoed audio
-p percentage changes the pitch of all audio echoed to a sound
card. Only the copy to the soundcard is affected, the recorded
audio samples in a file remain the same. Normal pitch, which is
the default, is given by 100%. Lower percentages correspond to
lower pitches, i.e. -p 50 transposes the audio output one
octave lower. See also the script pitchplay as an example. This
option was contributed by Raul Sobon.
Select mono or stereo recording
-m or -c 1 selects mono recording (both stereo channels are
mixed), -s or -c 2 or -c s selects stereo recording. Parameter s
will swap both sound channels.
Select maximum quality
-x will set stereo, 16 bits per sample at 44.1 KHz (full CD
quality). Note that other format options given later can change
Select sample quality
-b 8 specifies 8 bit (1 Byte) for each sample in each channel;
-b 12 specifies 12 bit (2 Byte) for each sample in each channel;
-b 16 specifies 16 bit (2 Byte) for each sample in each channel
(Ensure that your sample player or sound card is capable of
playing 12-bit or 16-bit samples). Selecting 12 or 16 bits dou-
bles file size. 12-bit samples are aligned to 16-bit samples,
so they waste some disk space.
Select sample rate
-r samplerate selects a sample rate. samplerate can be in a
range between 44100 and 900. Option -R lists all available
Select sample rate divider
-a divider selects a sample rate divider. divider can be mini-
mally 1 and maximally 50.5 and everything between in steps of
0.5. Option -R lists all available rates.
To make the sound smoother at lower sampling rates, cdda2wav
sums over n samples (where n is the specific dividend). So for
22050 Hertz output we have to sum over 2 samples, for 900 Hertz
we have to sum over 49 samples. This cancels higher frequen-
cies. Standard sector size of an audio CD (ignoring additional
information) is 2352 Bytes. In order to finish summing for an
output sample at sector boundaries the rates above have to be
choosen. Arbitrary sampling rates in high quality would require
some interpolation scheme, which needs much more sophisticated
List a table of all sampling rates
-R shows a list of all sample rates and their dividers. Dividers
can range from 1 to 50.5 in steps of 0.5.
Select start track and optionally end track
-t n+m selects n as the start track and optionally m as the last
track of a range to be recorded. These tracks must be from the
table of contents. This sets the track where recording begins.
Recording can advance through the following tracks as well (lim-
ited by the optional end track or otherwise depending on record-
ing time). Whether one file or different files are then created
depends on the -B option (see below).
Select start index
-i n selects the index to start recording with. Indices other
than 1 will invoke the index scanner, which will take some time
to find the correct start position. An offset may be given addi-
tionally (see below).
Set recording time
-d n sets recording time to n seconds or set recording time for
whole track if n is zero. In order to specify the duration in
frames (sectors) also, the argument can have an appended ’f’.
Then the numerical argument is to be taken as frames (sectors)
rather than seconds. Please note that if track ranges are being
used they define the recording time as well thus overriding any
-d option specified times.
Recording time is defined as the time the generated sample will
play (at the defined sample rate). Since it’s related to the
amount of generated samples, it’s not the time of the sampling
process itself (which can be less or more). It’s neither
strictly coupled with the time information on the audio CD
(shown by your hifi CD player). Differences can occur by the
usage of the -o option (see below). Notice that recording time
will be shortened, unless enough disk space exists. Recording
can be aborted at anytime by pressing the break character (sig-
Record all tracks of a complete audio CD in seperate files
-B copies each track into a seperate file. A base name can be
given. File names have an appended track number and an extension
corresponding to the audio format. To record all audio tracks of
a CD, use a sufficient high duration (i.e. -d99999).
Set start sector offset
-o sectors increments start sector of the track by sectors. By
this option you are able to skip a certain amount at the
beginning of a track so you can pick exactly the part you want.
Each sector runs for 1/75 seconds, so you have very fine con-
trol. If your offset is so high that it would not fit into the
current track, a warning message is issued and the offset is
ignored. Recording time is not reduced. (To skip introductory
quiet passages automagically, use the -w option see below.)
Wait for signal option
-w Turning on this option will suppress all silent output at
startup, reducing possibly file size. cdda2wav will watch for
any signal in the output signal and switches on writing to file.
Find extrem samples
-F Turning on this option will display the most negative and the
most positive sample value found during recording for both chan-
nels. This can be useful for readjusting the volume. The values
shown are not reset at track boundaries, they cover the complete
sampling process. They are taken from the original samples and
have the same format (i.e. they are independent of the selected
Find if input samples are in mono
-G If this option is given, input samples for both channels will
be compared. At the end of the program the result is printed.
Differences in the channels indicate stereo, otherwise when both
channels are equal it will indicate mono.
Undo the pre-emphasis in the input samples
-T Some older audio CDs are recorded with a modified frequency
response called pre-emphasis. This is found mostly in classical
recordings. The correction can be seen in the flags of the Table
Of Contents often. But there are recordings, that show this set-
ting only in the subchannels. If this option is given, the index
scanner will be started, which reads the q-subchannel of each
track. If pre-emphasis is indicated in the q-subchannel of a
track, but not in the TOC, pre-emphasis will be assumed to be
present, and subsequently a reverse filtering is done for this
track before the samples are written into the audio file.
Set audio format
-O audiotype can be wav (for wav files) or au or sun (for sun
PCM files) or cdr or raw (for headerless files to be used for cd
writers). All file samples are coded in linear pulse code modu-
lation (as done in the audio compact disc format). This holds
for all audio formats. Wav files are compatible to Wind*ws
sound files, they have lsb,msb byte order as being used on the
audio cd. The default filename extension is ’.wav’. Sun type
files are not like the older common logarithmically coded .au
files, but instead as mentioned above linear PCM is used. The
byte order is msb,lsb to be compatible. The default filename
extension is ’.au’. The AIFF and the newer variant AIFC from
the Apple/SGI world store their samples in bigendian format
(msb,lsb). In AIFC no compression is used. Finally the easiest
’format’, the cdr aka raw format. It is done per default in
msb,lsb byte order to satisfy the order wanted by most cd writ-
ers. Since there is no header information in this format, the
sample parameters can only be identified by playing the samples
on a soundcard or similiar. The default filename extension is
’.cdr’ or ’.raw’.
Select cdrom drive reading speed
-S speed allows to switch the cdrom drive to a certain level of
speed in order to reduce read errors. The argument is transfered
verbatim to the drive. Details depend very much on the cdrom
drives. An argument of 0 for example is often the default speed
of the drive, a value of 1 often selects single speed.
Enable MD5 checksums
-M count enables calculation of MD-5 checksum for ’count’ bytes
from the beginning of a track. This was introduced for quick
comparisons of tracks.
Use Monty’s libparanoia for reading of sectors
-paranoia selects an alternate way of extracting audio sectors.
Monty’s library is used with the following default options:
PARANOIA_MODE_FULL, but without PARANOIA_MODE_NEVERSKIP
for details see Monty’s libparanoia documentation. In this case
the option -P has no effect.
Do linear or overlapping reading of sectors
(This applies unless option -paranoia is used.) -P sectors
sets the given number of sectors for initial overlap sampling
for jitter correction. Two cases are to be distinguished. For
nonzero values, some sectors are read twice to enable cdda2wav’s
jitter correction. If an argument of zero is given, no overlap
sampling will be used. For nonzero overlap sectors cdda2wav
dynamically adjusts the setting during sampling (like cdparanoia
does). If no match can be found, cdda2wav retries the read with
an increased overlap. If the amount of jitter is lower than the
current overlapped samples, cdda2wav reduces the overlap set-
ting, resulting in a higher reading speed. The argument given
has to be lower than the total number of sectors per request
(see option -n below). Cdda2wav will check this setting and
issues a error message otherwise. The case of zero sectors is
nice on low load situations or errorfree (perfect) cdrom drives
and perfect (not scratched) audio cds.
Set the transfer size
-n sectors will set the transfer size to the specified sectors
Set number of ring buffer elements
-l buffers will allocate the specified number of ring buffer
Set endianess of input samples
-C endianess will override the default settings of the input
format. Endianess can be set explicitly to "little" or "big" or
to the automatic endianess detection based on voting with
Set endianess of output samples
-E endianess (endianess can be "little" or "big") will override
the default settings of the output format.
-v itemlist prints more information. A list allows selection of
different information items.
disable keeps quiet
toc displays the table of contents
summary displays a summary of recording parameters
indices invokes the index scanner and displays start positions
catalog retrieves and displays a media catalog number
trackid retrieves and displays international standard recording
sectors displays track start positions in absolute sector nota-
To combine several requests just list the suboptions seperated
The table of contents
The display will show the table of contents with number of
tracks and total time (displayed in mm:ss.hh format, mm=minutes,
ss=seconds, hh=rounded 1/100 seconds). The following list dis-
plays track number and track time for each entry. The summary
gives a line per track describing the type of the track.
track preemphasis copypermitted tracktype chans
The track column holds the track number. preemphasis shows if
that track has been given a non linear frequency response.
NOTE: You can undo this effect with the -T option. copy-permit-
ted indicates if this track is allowed to copy. tracktype can
be data or audio. On multimedia CDs (except hidden track CDs)
both of them should be present. channels is defined for audio
tracks only. There can be two or four channels.
No file output
-N this debugging option switches off writing to a file.
No infofile generation
-H this option switches off creation of an info file and a cddb
Generation of simple output for gui frontends
-g this option switches on simple line formatting, which is
needed to support gui frontends (like xcd-roast).
Verbose SCSI logging
-V this option switches on logging of SCSI commands. This will
produce a lot of output (when SCSI devices are being used).
This is needed for debugging purposes. The format is the same as
being used with the cdrecord program from Joerg Schilling. I
will not describe it here.
-q suppresses all screen output except error messages. That
reduces cpu time resources.
Just show information option
-J does not write a file, it only prints information about the
disc (depending on the -v option). This is just for information
Lookup album and track titles option
-L cddbp mode Cdda2wav tries to retrieve performer, album-, and
track titles from a cddbp server. The default server right now
is ’freedb.freedb.org’. It is planned to have more control over
the server handling later. The parameter defines how multiple
entries are handled:
0 interactive mode, the user chooses one of the entries.
1 take the first entry without asking.
Set server for title lookups
cddbp-server servername When using -L or --cddb, the server
being contacted can be set with this option.
Set portnumber for title lookups
cddbp-port portnumber When using -L or --cddb, the server port
being contacted can be set with this option.
HINTS ON USAGE
Don’t create samples you cannot read. First check your sample player
software and sound card hardware. I experienced problems with very low
sample rates (stereo <= 1575 Hz, mono <= 3675 Hz) when trying to play
them with standard WAV players for sound blaster (maybe they are not
legal in WAV format). Most CD-Writers insist on audio samples in a
bigendian format. Now cdda2wav supports the -E endianess option to
control the endianess of the written samples.
If your hardware is fast enough to run cdda2wav uninterrupted and your
CD drive is one of the ’perfect’ ones, you will gain speed when switch-
ing all overlap sampling off with the -P 0 option. Further fine tuning
can be done with the -n sectors option. You can specify how much sec-
tors should be requested in one go.
Cdda2wav supports pipes now. Use a filename of - to let cdda2wav output
its samples to standard output.
Conversion to other sound formats can be done using the sox program
package (although the use of sox -x to change the byte order of samples
should be no more necessary; see option -E to change the output byte-
If you want to sample more than one track into different files in one
run, this is currently possible with the -B option. When recording time
exceeds the track limit a new file will be opened for the next track.
Cdda2wav can generate a lot of files for various purposes.
There are audio files containing samples with default extensions These
files are not generated when option (-N) is given. Multiple files may
be written when the bulk copy option (-B) is used. Individual file
names can be given as arguments. If the number of file names given is
sufficient to cover all included audio tracks, the file names will be
used verbatim. Otherwise, if there are less file names than files
needed to write the included tracks, the part of the file name before
the extension is extended with ’_dd’ where dd represents the current
Cddb and Cdindex files:
If cdda2wav detects cd-extra or cd-text (album/track) title informa-
tion, then .cddb and .cdindex files are generated unless suppressed by
the option -H. They contain suitable formatted entries for submission
to audio cd track title databases in the internet. The CDINDEX and
CDDB(tm) systems are currently supported. For more information please
visit www.musicbrainz.org and www.freedb.com.
The inf files are describing the sample files and the part from the
audio cd, it was taken from. They are a means to transfer information
to a cd burning program like cdrecord. For example, if the original
audio cd had pre-emphasis enabled, and cdda2wav -T did remove the pre-
emphasis, then the inf file has pre-emphasis not set (since the audio
file does not have it anymore), while the .cddb and the .cdindex have
pre-emphasis set as the original does.
IMPORTANT: it is prohibited to sell copies of copyrighted material by
noncopyright holders. This program may not be used to circumvent copy-
rights. The user acknowledges this constraint when using the software.
Generation of md5 checksums is currently broken.
Performance may not be optimal on slower systems.
The index scanner may give timeouts.
The resampling (rate conversion code) uses polynomial interpolation,
which is not optimal.
Cdda2wav should use threads.
Cdda2wav currently cannot sample hidden audio tracks (track 1 index 0).
Thanks goto Project MODE (http://www.mode.net/) and Fraunhofer Institut
fuer integrierte Schaltungen (FhG-IIS) (http://www.iis.fhg.de/) for
financial support. Plextor Europe and Ricoh Japan provided cdrom disk
drives and cd burners which helped a lot to develop this software.
Rammi has helped a lot with the debugging and showed a lot of stamina
when hearing 100 times the first 16 seconds of the first track of the
Krupps CD. Libparanoia contributed by Monty (Christopher Montgomery)
Heiko Eissfeldt firstname.lastname@example.org
11 Sep 2002
Man(1) output converted with