For Programmers: Free Programming Magazines  


Home > Archive > Compression > November 2004 > Packet audio compression









You are viewing an archived Text-only version of the thread. To view this thread in it's original format and/or if you want to reply to this thread please [click here]

 

Author Packet audio compression
drk

2004-11-23, 3:55 pm

Hi,
i need an open source library for audio compression that work packet by
packet !!!

thank's in advance





Phil Frisbie, Jr.

2004-11-23, 3:55 pm

drk wrote:

> Hi,
> i need an open source library for audio compression that work packet by
> packet !!!


http://www.hawksoft.com/hawkvoice/

> thank's in advance



--
Phil Frisbie, Jr.
Hawk Software
http://www.hawksoft.com

Michel Bardiaux

2004-11-24, 3:55 pm

Phil Frisbie, Jr. wrote:
> drk wrote:
>
>
>
> http://www.hawksoft.com/hawkvoice/


From that page:

Kbps codec name license code type

64 G.711 u-law LGPL fixed point
32 Intel/DVI ADPCM Free * fixed point
13.2 GSM LGPL fixed point
4.8 LPC LGPL floating point
4.5-2.3 CELP LGPL floating point
2.4 LPC10 LGPL $ floating point
Up to 2.4 VBR-LPC10 LGPL $ floating point
1.8-1.4 OpenLPC Free *$ floating point

I suspect the OP when writing 'by packet' meant 'each packet can be
decoded independently of any other packet', i.e. a sateless decoder, aka
a stream with random access.

Now GSM is *definitely* stateful; ADPCM can be or not depending on file
format; I dont know the others enough to have an opinion. Only G711 is
*definitely* stateless.

The same remarks will apply to almost any opensource (or not) multimedia
library, like www.ffmpeg.org.
>
>
>
>



--
Michel Bardiaux
Peaktime Belgium S.A. Bd. du Souverain, 191 B-1160 Bruxelles
Tel : +32 2 790.29.41
Phil Frisbie, Jr.

2004-11-24, 3:55 pm

Michel Bardiaux wrote:

> Phil Frisbie, Jr. wrote:
>
>
>
> From that page:
>
> Kbps codec name license code type
>
> 64 G.711 u-law LGPL fixed point
> 32 Intel/DVI ADPCM Free * fixed point
> 13.2 GSM LGPL fixed point
> 4.8 LPC LGPL floating point
> 4.5-2.3 CELP LGPL floating point
> 2.4 LPC10 LGPL $ floating point
> Up to 2.4 VBR-LPC10 LGPL $ floating point
> 1.8-1.4 OpenLPC Free *$ floating point
>
> I suspect the OP when writing 'by packet' meant 'each packet can be
> decoded independently of any other packet', i.e. a sateless decoder, aka
> a stream with random access.


You may be right, and I thought about that possibility also, but he was not clear.

> Now GSM is *definitely* stateful; ADPCM can be or not depending on file
> format; I dont know the others enough to have an opinion. Only G711 is
> *definitely* stateless.


Yes, it is trivial to make ADPCM stateless, so he has at least two choices, if
that is what he really meant!
[color=darkred]
> The same remarks will apply to almost any opensource (or not) multimedia
> library, like www.ffmpeg.org.
>

--
Phil Frisbie, Jr.
Hawk Software
http://www.hawksoft.com

drk

2004-11-25, 3:55 pm

the hawkvoice codec is very fast an simply to use.
I have processing my data with every codec hawkvoice and i have compared
resultant
FTT using soundforge ; in conclusion only g726 5bit respect the original
spectrum the
result of other codec is not comparable.

I have and input signal 8khz 8bit and to use hawkvoice codecs i transform
it to 16 bit signed.

With g726 5bit my compression ratio is very low..


thank's in advance and excuse my for my english :)




"Phil Frisbie, Jr." <phil@hawksoft.com> ha scritto nel messaggio
news:kU3pd.7498$_3.92511@typhoon.sonic.net...
> Michel Bardiaux wrote:
>
>
> You may be right, and I thought about that possibility also, but he was

not clear.
>
>
> Yes, it is trivial to make ADPCM stateless, so he has at least two

choices, if
> that is what he really meant!
>
>
> --
> Phil Frisbie, Jr.
> Hawk Software
> http://www.hawksoft.com
>



Michel Bardiaux

2004-11-26, 3:55 pm

drk wrote:
> the hawkvoice codec is very fast an simply to use.
> I have processing my data with every codec hawkvoice and i have compared
> resultant
> FTT using soundforge ; in conclusion only g726 5bit respect the original
> spectrum the
> result of other codec is not comparable.
>
> I have and input signal 8khz 8bit and to use hawkvoice codecs i transform
> it to 16 bit signed.
>
> With g726 5bit my compression ratio is very low..
>
>
> thank's in advance and excuse my for my english :)
>
>


Is there a question relevant to this thread? Or are you just expressing
satisfaction with the hawkvoice codec? If so, please read carefully the
previous messages. The question was not whether it (or any other codec)
is good, but whether it is stateless.

>
>
> "Phil Frisbie, Jr." <phil@hawksoft.com> ha scritto nel messaggio
> news:kU3pd.7498$_3.92511@typhoon.sonic.net...
>
>
> not clear.
>
>
> choices, if
>
>
>
>



--
Michel Bardiaux
Peaktime Belgium S.A. Bd. du Souverain, 191 B-1160 Bruxelles
Tel : +32 2 790.29.41
Phil Frisbie, Jr.

2004-11-29, 3:56 pm

drk wrote:

> the hawkvoice codec is very fast an simply to use.
> I have processing my data with every codec hawkvoice and i have compared
> resultant
> FTT using soundforge ; in conclusion only g726 5bit respect the original
> spectrum the
> result of other codec is not comparable.


Yes, the ADPCM codec is a wave encoder, but the other lower bit rate codecs are
voice encoders. Voice encoders discard part of spectrum (lows and highs) before
encoding, and then the decoders synthesize the speech so that it resembles the
original, but the whole process is lossy.

> I have and input signal 8khz 8bit and to use hawkvoice codecs i transform
> it to 16 bit signed.
>
> With g726 5bit my compression ratio is very low..


Yes, I would only use it on a LAN when bandwidth is not a concern.

> thank's in advance and excuse my for my english :)


--
Phil Frisbie, Jr.
Hawk Software
http://www.hawksoft.com

Sponsored Links







Also available: Server administration forum archive | Web Design forum archive | Software forum archive | Hardware reviews archive

Copyright 2008 codecomments.com