Need expert in live audio restoration on-the-fly

[darius2]

Hi,

need expert in live audio restoration on-the-fly.

Please tell me if Audacity or any other audio processor like GoldWave, Dart Pro,
can process live audio stream on-the-fly. restoring, repairing signal amplitude drops (to zero)
by replacing missing audio signal intervals with Copy&Paste preceding audio track.

I am looking for the following feature:
soundcard to soundcard real-time processing.

Tell me what settings, feature in Audacity is fit for removing hiss, crackle, noise, and buzz from original audio track.

[kozikowski]

Audacity doesn't do any processing in real time.
Koz

[darius2]

Ok.
I can start in off-line mode.
But what can be done with signal drops ?

[steve]

But what can be done with signal drops ?

You mean when chunks of data are missing? (drop-outs)
There are some techniques for concealing very small amounts of data loss (packet loss concealment), but if the lost data is much more than loosing the occasional few bytes there's not much that can be done about it. Theoretically it should be possible to implement something like the Audacity "Repair Tool" in (almost) real time, but I'm not aware of any programs that have that feature - perhaps some specialist broadcasting/telecommunications software? Generally in audio applications the approach is to try and avoid data loss rather than to try and patch it up.
There's some information about packet loss concealment on the Speex website: http://www.speex.org/

[darius2]

But what can be done with signal drops ?

You mean when chunks of data are missing? (drop-outs)
There are some techniques for concealing very small amounts of data loss (packet loss concealment), but if the lost data is much more than loosing the occasional few bytes there's not much that can be done about it. Theoretically it should be possible to implement something like the Audacity "Repair Tool" in (almost) real time, but I'm not aware of any programs that have that feature - perhaps some specialist broadcasting/telecommunications software? Generally in audio applications the approach is to try and avoid data loss rather than to try and patch it up.
There's some information about packet loss concealment on the Speex website: http://www.speex.org/

Exactly, drop-outs.
As lost data is more than few bytes, Audacity " Repair Tool" is no use as limited up to 128 samples.

Dart Pro demo is coming with limited set of features, but is marketed as real-time filter.

No entry for "Packet loss concealment" in manual to speex (PDF).

Google search for "Packet loss concealment" is exactly the right way

http://www.google.com/#hl=en&source=hp& ... be010d257f


PLC Wiki
http://en.wikipedia.org/wiki/Packet_loss_concealment


good examples

Silence Insertion Replay last packet G.711 Appendix 1
http://www.voiptroubleshooter.com/problems/plc.html

I would like to implement all of above 3 adaptive PLC algorithms/ filters into my sound card driver for Vista, if available.


Is it possible to enable PLC or adaptive PLC for Astersk ?

-----
btw

I have installed Audacity 1.3 and old problem with saving audio as MusicOnHold mono, 16 bit, 8000 Hz PCM wav.

I am saving wav file as 8000 Hz, as set up in Preferences / Quality / Sampling
and on opening get file resampled to 44100 Hz 75 kb/s

OK I set project rate to 8000 Hz
and now can open wav file as 8000 Hz
but on opening a saved file, calculating waveform message is displayed
and it takes 10-20 s to open 7 s wav file.

At the same time, old stereo 44100 Hz 16 bit gets opened immediately, so it looks like 8000 Hz is not a default project rate if selected so
and Audacity 1.3 keeps doing some resampling (saved file is 128 kb/s anyway).

Any option available to see original file info, details like rate, 16 bit, mono/stereo from within Audacity ?

[darius2]

follow-up

re. speex
http://astbook.asteriskdocs.org/en/2nd_ ... CT-13.html

"
The codecs.conf file is fairly new in Asterisk, and as of this writing it allows configuration of Speex parameters only. The settings are self-explanatory, as long as you are familiar with the Speex protocol (see http://www.speex.org).

codecs.conf also allows you to configure Packet Loss Concealment (PLC). You need to define a [plc] section and indicate genericplc => true. This will cause Asterisk to attempt to interpolate any packets that are missed. (Enabling this functionality will incur a small performance penalty.)

"

Ok, my codecs.conf file is coming both with [speex] and [plc} contexts
genericplc => true (so enabled)

[speex] context is coming with a number of parameters, I can set up

But it looks to me, problem is not with Asterisk + jitterbuffer, as Asterisk is playing MOH files fine, when accessed from a SIP softphone, run on the same laptop.
The problem looks to be related to celliax audio channel, middle-ware connecting Asterisk to a sound card, so audio files can be played directly to headphone.


follow-up

Just downloaded and installed Speex for Windows

http://www.roed.republika.pl/speexw/

DOS app
as input should standard spx file input_file.spx or sdtin -
and output can be wav or stdout

Any chance to have speex to process live audio stream in real-time ?

In my case audio stream comes from a sound card, should be Packet Loss Concealment (PLC) processed by speex and sent to stdout, so to headphone
so it looks problematic to have Speex to work with one sound card installed only,

Any idea ?

[Gale Andrews]

I have installed Audacity 1.3 and old problem with saving audio as MusicOnHold mono, 16 bit, 8000 Hz PCM wav.

I am saving wav file as 8000 Hz, as set up in Preferences / Quality / Sampling
and on opening get file resampled to 44100 Hz 75 kb/s

OK I set project rate to 8000 Hz
and now can open wav file as 8000 Hz
but on opening a saved file, calculating waveform message is displayed
and it takes 10-20 s to open 7 s wav file.

At the same time, old stereo 44100 Hz 16 bit gets opened immediately, so it looks like 8000 Hz is not a default project rate if selected so and Audacity 1.3 keeps doing some resampling (saved file is 128 kb/s anyway).

If the file you are importing is 44100 Hz and you import it into an empty project with Quality Preferences set to a default sample rate of 8000 Hz (or to any rate other than 44100 Hz), then the project rate will change to 44100 Hz to reflect the rate of the imported file. That behaviour isn't configurable but you can add your vote for having a preference to always import files at the default sample rate.

Any option available to see original file info, details like rate, 16 bit, mono/stereo from within Audacity ?

Obviously you can see whether the imported file is mono or stereo, otherwise, no. Would you like to make that a feature request too? The rate given above the mute/solo buttons is always the current rate of the track in Audacity, though until you resample it, that will correspond with the rate of the original file. The bit-rate given is also the current resolution, so given the default sample format is 32-bit float, that won't match with the resolution of the original file unless it was a 32-bit float file.

There are lots of apps that give you codec, rate and format size information, such as MediaInfo.



Gale

[darius2]

If the file you are importing is 44100 Hz and you import it into an empty project with Quality Preferences set to a default sample rate of 8000 Hz (or to any rate other than 44100 Hz), then the project rate will change to 44100 Hz to reflect the rate of the imported file. That behaviour isn't configurable but you can add your vote for having a preference to always import files at the default sample rate.

Exactly, any preset Project rate in Hz is reset by Hz rate of an opened file.
As I don't save projects but PCM wav files only,
I am not sure what are default or preset settings for PCM wav audio file export.

Any option available to see original file info, details like rate, 16 bit, mono/stereo from within Audacity ?

Obviously you can see whether the imported file is mono or stereo, otherwise, no. Would you like to make that a feature request too? The rate given above the mute/solo buttons is always the current rate of the track in Audacity, though until you resample it, that will correspond with the rate of the original file. The bit-rate given is also the current resolution, so given the default sample format is 32-bit float, that won't match with the resolution of the original file unless it was a 32-bit float file.

There are lots of apps that give you codec, rate and format size information, such as MediaInfo. Gale

Is wav file already opened, any change in Project rate Hz changes nothing (immediate change)


oK.

1. wav file is opened with its native parameters

2. Project rate Hz is reset to the rate of opened file ?
Project rate can be subsequently changed to any preset Hz value, not resulting in any action
3. Audacity comes with default settings - Preferences

4. Rate and format can be temporary set for an already opened file

5. File can be exported -as is- ?
import/export settings don't influence parameters of exported file

6. Exported file can be saved on export and new bitrate, mono/stereo, parameters can be selected

Isn't it too much for a file to be exported/saved ?

as export/save parameters can be selected, set from 4 seperated windows ( as above).