[Freeswitch-users] .wav vs .gsm file sizes for recording calls.

Discussion:

Wes

2011-06-24 21:26:06 UTC

In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.

I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?

checking the file type using "file" I see that the larger one is:
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz

and the wav created by sox via the default conversion from .gsm is:
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz

So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...

David Ponzone

2011-06-24 21:41:51 UTC

Permalink

Wes,

it's because you are confusing 2 things.
GSM is a codec, so it's a sound format.
WAVE is a kind of container (like MKV) that can contain various codecs.
In your case, you probably did a mistake using sox (it's not an easy tool), so you ended up with a wav which is still GSM audio inside.

You have to tell sox which exact output format you want.
AFAIR, you have to use the -s option (for signed linear).

David Ponzone Direction Technique
email: david.ponzone at ipeva.fr
tel: 01 74 03 18 97
gsm: 06 66 98 76 34

Service Client IPeva
tel: 0811 46 26 26
www.ipeva.fr - www.ipeva-studio.com

Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis ? l'intention exclusive de ses destinataires. Toute utilisation ou diffusion non autoris?e est interdite. Tout message ?lectronique est susceptible d'alt?ration. IPeva d?cline toute responsabilit? au titre de ce message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous n'?tes pas destinataire de ce message, merci de le d?truire imm?diatement et d'avertir l'exp?diteur.

Post by Wes
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110624/07f35c08/attachment.html

Jan Berger

2011-06-24 21:53:51 UTC

Permalink

A gentle warning when using GSM codec?s

Every codec compress voice at the cost of quality. And some codec?s like GSM
and G.729 are not a nice combination. If you use GSM to save disk space and
play it back using G.729 the resulting voice might contain some added noise
due to algorithmic incompatibility.

jan

_____

From: freeswitch-users-bounces at lists.freeswitch.org
[mailto:freeswitch-users-bounces at lists.freeswitch.org] On Behalf Of David
Ponzone
Sent: 24. juni 2011 23:42
To: FreeSWITCH Users Help
Subject: Re: [Freeswitch-users] .wav vs .gsm file sizes for recording calls.

Wes,

it's because you are confusing 2 things.

GSM is a codec, so it's a sound format.

WAVE is a kind of container (like MKV) that can contain various codecs.

In your case, you probably did a mistake using sox (it's not an easy tool),
so you ended up with a wav which is still GSM audio inside.

You have to tell sox which exact output format you want.

AFAIR, you have to use the -s option (for signed linear).

David Ponzone Direction Technique

email: david.ponzone at ipeva.fr

tel: 01 74 03 18 97

gsm: 06 66 98 76 34

Service Client IPeva

tel: 0811 46 26 26

www.ipeva.fr <BLOCKED::http://www.ipeva.fr/> - www.ipeva-studio.com
<BLOCKED::http://www.ipeva-studio.com/>

Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis ?
l'intention exclusive de ses destinataires. Toute utilisation ou diffusion
non autoris?e est interdite. Tout message ?lectronique est susceptible
d'alt?ration. IPeva d?cline toute responsabilit? au titre de ce message s'il
a ?t? alt?r?, d?form? ou falsifi?. Si vous n'?tes pas destinataire de ce
message, merci de le d?truire imm?diatement et d'avertir l'exp?diteur.

Le 24/06/2011 ? 23:26, Wes a ?crit :

In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.

I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?

checking the file type using "file" I see that the larger one is:
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz

and the wav created by sox via the default conversion from .gsm is:
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz

So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...

_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE

FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110624/b8c0cc44/attachment-0001.html

Steve Underwood

2011-06-25 03:11:42 UTC

Permalink

Practically any conversion of a low bit rate codec to another low bit
rate codec gives horrible results. Even low bit rate codec -> linear ->
the same low bit rate codec can be pretty horrible, unless the two
compression steps are synchronised.

A lot of the complaints you will find about the horrible sound from some
codec has little to do with the codec's performance. Someone has tried
the codec with audio that had already been compressed by some other low
bit rate codec.

Steve

Post by Jan Berger
A gentle warning when using GSM codec?s
Every codec compress voice at the cost of quality. And some codec?s
like GSM and G.729 are not a nice combination. If you use GSM to save
disk space and play it back using G.729 the resulting voice might
contain some added noise due to algorithmic incompatibility.
jan
------------------------------------------------------------------------
*From:*freeswitch-users-bounces at lists.freeswitch.org
[mailto:freeswitch-users-bounces at lists.freeswitch.org] *On Behalf Of
*David Ponzone
*Sent:* 24. juni 2011 23:42
*To:* FreeSWITCH Users Help
*Subject:* Re: [Freeswitch-users] .wav vs .gsm file sizes for
recording calls.
Wes,
it's because you are confusing 2 things.
GSM is a codec, so it's a sound format.
WAVE is a kind of container (like MKV) that can contain various codecs.
In your case, you probably did a mistake using sox (it's not an easy
tool), so you ended up with a wav which is still GSM audio inside.
You have to tell sox which exact output format you want.
AFAIR, you have to use the -s option (for signed linear).
David Ponzone Direction Technique
email: david.ponzone at ipeva.fr <mailto:david.ponzone at ipeva.fr>
tel: 01 74 03 18 97
gsm: 06 66 98 76 34
Service ClientIPeva
tel: 0811 46 26 26
_www.ipeva.fr <BLOCKED::http://www.ipeva.fr/>_- _www.ipeva-studio.com
<BLOCKED::http://www.ipeva-studio.com/>_
/Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis
? l'intention exclusive de ses destinataires. Toute utilisation ou
diffusion non autoris?e est interdite. Tout message ?lectronique est
susceptible d'alt?ration. *IPeva* d?cline toute responsabilit? au
titre de ce message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous
n'?tes pas destinataire de ce message, merci de le d?truire
imm?diatement et d'avertir l'exp?diteur./
__
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
<mailto:FreeSWITCH-users at lists.freeswitch.org>
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

Jan Berger

2011-06-24 21:46:00 UTC

Permalink

Wav is just a standard file format, the size is decided by the contents.

The first format is PCM which is 8000 bytes samples per sec
The second is GSM which is compressed voice - so in short you changed the
file format, but is was and still is GSM inside.

To convert from PCM to GSM you need an application that can do so - WavePad
etc. can be recommended.

Jan

-----Original Message-----
From: freeswitch-users-bounces at lists.freeswitch.org
[mailto:freeswitch-users-bounces at lists.freeswitch.org] On Behalf Of Wes
Sent: 24. juni 2011 23:26
To: FreeSWITCH Users Help
Subject: [Freeswitch-users] .wav vs .gsm file sizes for recording calls.

In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.

I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?

checking the file type using "file" I see that the larger one is:
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz

and the wav created by sox via the default conversion from .gsm is:
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz

So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...

_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE

FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

Michael Collins

2011-06-24 23:31:14 UTC

Permalink

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110624/ad6e1a70/attachment-0001.html

Wes

2011-06-27 19:43:38 UTC

Permalink

I guess part of my confusion here was due to the term "raw data"
mentioned in conjunction with the .gsm extension on the wiki page
below... but actually gsm is a compressed format.

http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session

So, what is the best "compressed" format to use for recording voice
(that is available as a direct recording format inside freeswitch)?
There are tons of formats listed when I do "show file", but I tried a
few and they are also giving me large files like the wav extension did.
(au, for example)

Even though the PCM/Wave format is preferred for voice quality, when
we're talking about a 10:1 compression ratio, if the sound quality is
still acceptable, I'd rather just record directly into the compressed
format. We're talking about ~10- 20 minute recordings that will need to
be transferred over the internet to a third party.

Post by Michael Collins
I would caution you to consider adding disk space before you try to
compress all your recordings. The 16 bit SLIN that FS normally puts in
your wave files are pretty easy to handle, whether playing back in a
FS session, or encoding for playback on some other device.
An alternative might be to use lame to convert them to MP3's or
ogg/vorbis files. If you look on the main FS conf call page you'll see
I have the weekly recordings in multiple formats.
(http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls)
Here are some stats for last Wednesday's call. Note that I record wave
files in 48kHz then use sox to downsample to 16kHz wave, then I
convert that 16kHz file into MP3 and Vorbis (in an ogg container).
<2831>:ls -1s conf_call_2011-06-15.*
18736 conf_call_2011-06-15.mp3
23044 conf_call_2011-06-15.ogg
199756 conf_call_2011-06-15.wav
<2832>:file conf_call_2011-06-15.mp3
conf_call_2011-06-15.mp3: MPEG ADTS, layer III, v2, 24 kBits, 16 kHz,
Monaural
<2833>:file conf_call_2011-06-15.ogg
conf_call_2011-06-15.ogg: Ogg data, Vorbis audio, mono, 16000 Hz,
~48000 bps, created by: Xiph.Org libVorbis I
<2834>:file conf_call_2011-06-15.wav
conf_call_2011-06-15.wav: RIFF (little-endian) data, WAVE audio,
Microsoft PCM, 16 bit, mono 16000 Hz
Note that the file sizes are in 1K blocks.
So, bottom line is this: if you have the disk space then use wave. If
you don't have disk space for wave then get some! :D If you REALLY
need to use a different format then choose something like MP3 or
Vorbis for long-term storage.
-MC
On Fri, Jun 24, 2011 at 2:26 PM, Wes <wes-fs at 499x.com
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
<mailto:FreeSWITCH-users at lists.freeswitch.org>
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110627/28ca525c/attachment.html

David Ponzone

2011-06-27 20:29:54 UTC

Permalink

You should really consider recording in SLIN, and then, batch compressing that later (during off-peak hours) to MP3 if you really need it to be smaller.
Any efficient compression takes CPU, and you may want to avoid doing that live on your server.

David Ponzone Direction Technique
email: david.ponzone at ipeva.fr
tel: 01 74 03 18 97
gsm: 06 66 98 76 34

Service Client IPeva
tel: 0811 46 26 26
www.ipeva.fr - www.ipeva-studio.com

Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis ? l'intention exclusive de ses destinataires. Toute utilisation ou diffusion non autoris?e est interdite. Tout message ?lectronique est susceptible d'alt?ration. IPeva d?cline toute responsabilit? au titre de ce message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous n'?tes pas destinataire de ce message, merci de le d?truire imm?diatement et d'avertir l'exp?diteur.

I guess part of my confusion here was due to the term "raw data" mentioned in conjunction with the .gsm extension on the wiki page below... but actually gsm is a compressed format.
http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session
So, what is the best "compressed" format to use for recording voice (that is available as a direct recording format inside freeswitch)? There are tons of formats listed when I do "show file", but I tried a few and they are also giving me large files like the wav extension did. (au, for example)
Even though the PCM/Wave format is preferred for voice quality, when we're talking about a 10:1 compression ratio, if the sound quality is still acceptable, I'd rather just record directly into the compressed format. We're talking about ~10- 20 minute recordings that will need to be transferred over the internet to a third party.

I would caution you to consider adding disk space before you try to compress all your recordings. The 16 bit SLIN that FS normally puts in your wave files are pretty easy to handle, whether playing back in a FS session, or encoding for playback on some other device.
An alternative might be to use lame to convert them to MP3's or ogg/vorbis files. If you look on the main FS conf call page you'll see I have the weekly recordings in multiple formats. (http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls)
<2831>:ls -1s conf_call_2011-06-15.*
18736 conf_call_2011-06-15.mp3
23044 conf_call_2011-06-15.ogg
199756 conf_call_2011-06-15.wav
<2832>:file conf_call_2011-06-15.mp3
conf_call_2011-06-15.mp3: MPEG ADTS, layer III, v2, 24 kBits, 16 kHz, Monaural
<2833>:file conf_call_2011-06-15.ogg
conf_call_2011-06-15.ogg: Ogg data, Vorbis audio, mono, 16000 Hz, ~48000 bps, created by: Xiph.Org libVorbis I
<2834>:file conf_call_2011-06-15.wav
conf_call_2011-06-15.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 16000 Hz
Note that the file sizes are in 1K blocks.
So, bottom line is this: if you have the disk space then use wave. If you don't have disk space for wave then get some! :D If you REALLY need to use a different format then choose something like MP3 or Vorbis for long-term storage.
-MC
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110627/8bb73322/attachment-0001.html

Wes

2011-06-27 20:45:44 UTC

Permalink

I'm sorry, but what is the SLIN format? I tried that extension and it's
invalid. I searched the wiki and found nothing.

Post by David Ponzone
You should really consider recording in SLIN, and then, batch
compressing that later (during off-peak hours) to MP3 if you really
need it to be smaller.
Any efficient compression takes CPU, and you may want to avoid doing
that live on your server.
David Ponzone Direction Technique
email: david.ponzone at ipeva.fr <mailto:david.ponzone at ipeva.fr>
tel: 01 74 03 18 97
gsm: 06 66 98 76 34
Service ClientIPeva
tel: 0811 46 26 26
www.ipeva.fr <BLOCKED::http://www.ipeva.fr/> - www.ipeva-studio.com
<BLOCKED::http://www.ipeva-studio.com/>
/Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis
? l'intention exclusive de ses destinataires. Toute utilisation ou
diffusion non autoris?e est interdite. Tout message ?lectronique est
susceptible d'alt?ration. /*/IPeva/*/ d?cline toute responsabilit? au
titre de ce message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous
n'?tes pas destinataire de ce message, merci de le d?truire
imm?diatement et d'avertir l'exp?diteur./
/
/

Post by Wes
I guess part of my confusion here was due to the term "raw data"
mentioned in conjunction with the .gsm extension on the wiki page
below... but actually gsm is a compressed format.
http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session
So, what is the best "compressed" format to use for recording voice
(that is available as a direct recording format inside freeswitch)?
There are tons of formats listed when I do "show file", but I tried a
few and they are also giving me large files like the wav extension
did. (au, for example)
Even though the PCM/Wave format is preferred for voice quality, when
we're talking about a 10:1 compression ratio, if the sound quality is
still acceptable, I'd rather just record directly into the compressed
format. We're talking about ~10- 20 minute recordings that will need
to be transferred over the internet to a third party.

Post by Michael Collins
I would caution you to consider adding disk space before you try to
compress all your recordings. The 16 bit SLIN that FS normally puts
in your wave files are pretty easy to handle, whether playing back
in a FS session, or encoding for playback on some other device.
An alternative might be to use lame to convert them to MP3's or
ogg/vorbis files. If you look on the main FS conf call page you'll
see I have the weekly recordings in multiple formats.
(http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls)
Here are some stats for last Wednesday's call. Note that I record
wave files in 48kHz then use sox to downsample to 16kHz wave, then I
convert that 16kHz file into MP3 and Vorbis (in an ogg container).
<2831>:ls -1s conf_call_2011-06-15.*
18736 conf_call_2011-06-15.mp3
23044 conf_call_2011-06-15.ogg
199756 conf_call_2011-06-15.wav
<2832>:file conf_call_2011-06-15.mp3
conf_call_2011-06-15.mp3: MPEG ADTS, layer III, v2, 24 kBits, 16 kHz, Monaural
<2833>:file conf_call_2011-06-15.ogg
conf_call_2011-06-15.ogg: Ogg data, Vorbis audio, mono, 16000 Hz,
~48000 bps, created by: Xiph.Org <http://Xiph.Org> libVorbis I
<2834>:file conf_call_2011-06-15.wav
conf_call_2011-06-15.wav: RIFF (little-endian) data, WAVE audio,
Microsoft PCM, 16 bit, mono 16000 Hz
Note that the file sizes are in 1K blocks.
So, bottom line is this: if you have the disk space then use wave.
If you don't have disk space for wave then get some! :D If you
REALLY need to use a different format then choose something like MP3
or Vorbis for long-term storage.
-MC
On Fri, Jun 24, 2011 at 2:26 PM, Wes <wes-fs at 499x.com
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com <http://www.cluecon.com/> 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
<mailto:FreeSWITCH-users at lists.freeswitch.org>
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org <http://www.freeswitch.org/>
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
<mailto:FreeSWITCH-users at lists.freeswitch.org>
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110627/5ad86d77/attachment-0001.html

Michael Collins

2011-06-27 23:42:16 UTC

Permalink

Post by Wes
**
I'm sorry, but what is the SLIN format? I tried that extension and it's
invalid. I searched the wiki and found nothing.

http://en.wikipedia.org/wiki/Linear_pulse-code_modulation

There's lots of names for it. Just think of SLIN as "the format normally
stored in a .wav file." (That's an imperfect definition but it will help you
get over the mental hump.)

In summary, just record to .wav files and x-code later when there are more
system resources available. You could also use "nice" or something like that
to prevent the x-coding process from grabbing all your system resources.

-MC

Post by Wes
You should really consider recording in SLIN, and then, batch compressing
that later (during off-peak hours) to MP3 if you really need it to be
smaller.
Any efficient compression takes CPU, and you may want to avoid doing that
live on your server.
David Ponzone Direction Technique
email: david.ponzone at ipeva.fr
tel: 01 74 03 18 97
gsm: 06 66 98 76 34
Service Client IPeva
tel: 0811 46 26 26
www.ipeva.fr - www.ipeva-studio.com
*Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis ?
l'intention exclusive de ses destinataires. Toute utilisation ou diffusion
non autoris?e est interdite. Tout message ?lectronique est susceptible
d'alt?ration. **IPeva** d?cline toute responsabilit? au titre de ce
message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous n'?tes pas
destinataire de ce message, merci de le d?truire imm?diatement et d'avertir
l'exp?diteur.*
*
*
I guess part of my confusion here was due to the term "raw data"
mentioned in conjunction with the .gsm extension on the wiki page below...
but actually gsm is a compressed format.
http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session
So, what is the best "compressed" format to use for recording voice (that
is available as a direct recording format inside freeswitch)? There are tons
of formats listed when I do "show file", but I tried a few and they are also
giving me large files like the wav extension did. (au, for example)
Even though the PCM/Wave format is preferred for voice quality, when we're
talking about a 10:1 compression ratio, if the sound quality is still
acceptable, I'd rather just record directly into the compressed format.
We're talking about ~10- 20 minute recordings that will need to be
transferred over the internet to a third party.
I would caution you to consider adding disk space before you try to
compress all your recordings. The 16 bit SLIN that FS normally puts in your
wave files are pretty easy to handle, whether playing back in a FS session,
or encoding for playback on some other device.
An alternative might be to use lame to convert them to MP3's or ogg/vorbis
files. If you look on the main FS conf call page you'll see I have the
weekly recordings in multiple formats. (
http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls)
Here are some stats for last Wednesday's call. Note that I record wave
files in 48kHz then use sox to downsample to 16kHz wave, then I convert that
16kHz file into MP3 and Vorbis (in an ogg container). Here's what the
<2831>:ls -1s conf_call_2011-06-15.*
18736 conf_call_2011-06-15.mp3
23044 conf_call_2011-06-15.ogg
199756 conf_call_2011-06-15.wav
<2832>:file conf_call_2011-06-15.mp3
conf_call_2011-06-15.mp3: MPEG ADTS, layer III, v2, 24 kBits, 16 kHz, Monaural
<2833>:file conf_call_2011-06-15.ogg
conf_call_2011-06-15.ogg: Ogg data, Vorbis audio, mono, 16000 Hz, ~48000
bps, created by: Xiph.Org libVorbis I
<2834>:file conf_call_2011-06-15.wav
conf_call_2011-06-15.wav: RIFF (little-endian) data, WAVE audio, Microsoft
PCM, 16 bit, mono 16000 Hz
Note that the file sizes are in 1K blocks.
So, bottom line is this: if you have the disk space then use wave. If you
don't have disk space for wave then get some! :D If you REALLY need to use a
different format then choose something like MP3 or Vorbis for long-term
storage.
-MC

_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicagohttp://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing listFreeSWITCH-users at lists.freeswitch.orghttp://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-usershttp://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicagohttp://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing listFreeSWITCH-users at lists.freeswitch.orghttp://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-usershttp://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110627/58d20e46/attachment-0001.html

Steven Ayre

2011-06-27 22:10:32 UTC

Permalink

Its not just voice quality. It's also CPU.

If you're using a compressed format, you're doing transcoding step. FS needs to decompress the file to L16 so it can compress it to the codec. Storing as wav containing l16 would avoid that. You can also use mod_native_file to compress the audio file to each codec format so that you don't need to transcode at all at the time because it's already done, although that does mean a file for each codec so more space needed.

Steve on iPhone

I guess part of my confusion here was due to the term "raw data" mentioned in conjunction with the .gsm extension on the wiki page below... but actually gsm is a compressed format.
http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session
So, what is the best "compressed" format to use for recording voice (that is available as a direct recording format inside freeswitch)? There are tons of formats listed when I do "show file", but I tried a few and they are also giving me large files like the wav extension did. (au, for example)
Even though the PCM/Wave format is preferred for voice quality, when we're talking about a 10:1 compression ratio, if the sound quality is still acceptable, I'd rather just record directly into the compressed format. We're talking about ~10- 20 minute recordings that will need to be transferred over the internet to a third party.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110627/48b7c5cb/attachment-0001.html

David Ponzone

2011-06-27 23:28:32 UTC

Permalink

This post might be inappropriate. Click to display it.

Steven Ayre

2011-06-28 07:58:10 UTC

Permalink

You're right I think. :) Was talking about playback... guess it was pretty
late.

-Steve

Post by David Ponzone
Steven,
AFAIR, one of the recording app (not sure which) does not support
mod_native_file, meaning for instance you can't dump your call using codec
XXX to a raw file.
That's a quite old limitation, so perhaps it's not an issue anymore.
David Ponzone Direction Technique
email: david.ponzone at ipeva.fr
tel: 01 74 03 18 97
gsm: 06 66 98 76 34
Service Client IPeva
tel: 0811 46 26 26
www.ipeva.fr - www.ipeva-studio.com
*Ce message et toutes les pi?ces jointes sont confidentiels et ?tablis ?
l'intention exclusive de ses destinataires. Toute utilisation ou diffusion
non autoris?e est interdite. Tout message ?lectronique est susceptible
d'alt?ration. **IPeva** d?cline toute responsabilit? au titre de ce
message s'il a ?t? alt?r?, d?form? ou falsifi?. Si vous n'?tes pas
destinataire de ce message, merci de le d?truire imm?diatement et d'avertir
l'exp?diteur.*
*
*
Its not just voice quality. It's also CPU.
If you're using a compressed format, you're doing transcoding step. FS
needs to decompress the file to L16 so it can compress it to the codec.
Storing as wav containing l16 would avoid that. You can also use
mod_native_file to compress the audio file to each codec format so that you
don't need to transcode at all at the time because it's already done,
although that does mean a file for each codec so more space needed.
Steve on iPhone
I guess part of my confusion here was due to the term "raw data" mentioned
in conjunction with the .gsm extension on the wiki page below... but
actually gsm is a compressed format.
<http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session>
http://wiki.freeswitch.org/wiki/Misc._Dialplan_Tools_record_session
So, what is the best "compressed" format to use for recording voice (that
is available as a direct recording format inside freeswitch)? There are tons
of formats listed when I do "show file", but I tried a few and they are also
giving me large files like the wav extension did. (au, for example)
Even though the PCM/Wave format is preferred for voice quality, when we're
talking about a 10:1 compression ratio, if the sound quality is still
acceptable, I'd rather just record directly into the compressed format.
We're talking about ~10- 20 minute recordings that will need to be
transferred over the internet to a third party.
I would caution you to consider adding disk space before you try to
compress all your recordings. The 16 bit SLIN that FS normally puts in your
wave files are pretty easy to handle, whether playing back in a FS session,
or encoding for playback on some other device.
An alternative might be to use lame to convert them to MP3's or ogg/vorbis
files. If you look on the main FS conf call page you'll see I have the
weekly recordings in multiple formats. (<http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls>
http://wiki.freeswitch.org/wiki/Weekly_Conference_Call#Past_Calls)
Here are some stats for last Wednesday's call. Note that I record wave
files in 48kHz then use sox to downsample to 16kHz wave, then I convert that
16kHz file into MP3 and Vorbis (in an ogg container). Here's what the
<2831>:ls -1s conf_call_2011-06-15.*
18736 conf_call_2011-06-15.mp3
23044 conf_call_2011-06-15.ogg
199756 conf_call_2011-06-15.wav
<2832>:file conf_call_2011-06-15.mp3
conf_call_2011-06-15.mp3: MPEG ADTS, layer III, v2, 24 kBits, 16 kHz, Monaural
<2833>:file conf_call_2011-06-15.ogg
conf_call_2011-06-15.ogg: Ogg data, Vorbis audio, mono, 16000 Hz, ~48000
bps, created by: Xiph.Org libVorbis I
<2834>:file conf_call_2011-06-15.wav
conf_call_2011-06-15.wav: RIFF (little-endian) data, WAVE audio, Microsoft
PCM, 16 bit, mono 16000 Hz
Note that the file sizes are in 1K blocks.
So, bottom line is this: if you have the disk space then use wave. If you
don't have disk space for wave then get some! :D If you REALLY need to use a
different format then choose something like MP3 or Vorbis for long-term
storage.
-MC

Post by Wes
In my tests, if I record a call in .wav format, a 10 second file is
about 177,000 bytes, while a 10 second .gsm file is 17,000 bytes.
I then used sox to convert the .gsm file to a .wav file, and it stayed
at around 17,000 bytes. So, is the default recording format for .wav
using a higher sample rate? vs the default conversion format for the sox
tool?
RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz
RIFF (little-endian) data, WAVE audio, GSM 6.10, mono 8000 Hz
So apparently the larger wav file is 16 bit... how are these recording
parameters controlled? Can I set it to record directly into the smaller
wav format? Or will I have to run sox on every file...
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
<http://www.cluecon.com/>http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
<FreeSWITCH-users at lists.freeswitch.org>
FreeSWITCH-users at lists.freeswitch.org
<http://lists.freeswitch.org/mailman/listinfo/freeswitch-users>
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:<http://lists.freeswitch.org/mailman/options/freeswitch-users>
http://lists.freeswitch.org/mailman/options/freeswitch-users
<http://www.freeswitch.org/>http://www.freeswitch.org

_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago <http://www.cluecon.com/>http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list <FreeSWITCH-users at lists.freeswitch.org>FreeSWITCH-users at lists.freeswitch.org <http://lists.freeswitch.org/mailman/listinfo/freeswitch-users>http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE: <http://lists.freeswitch.org/mailman/options/freeswitch-users>http://lists.freeswitch.org/mailman/options/freeswitch-users <http://www.freeswitch.org/>http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org
_______________________________________________
Join us at ClueCon 2011, Aug 9-11, Chicago
http://www.cluecon.com 877-7-4ACLUE
FreeSWITCH-users mailing list
FreeSWITCH-users at lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20110628/2c795e27/attachment-0001.html