sox-users Mailing List for SoX - Sound eXchange

Hello, i would like to know if there is a way to get the timestamps of
silence on a file without splitting the file. Or is there any other command
line utility that can do this besides sox?





--
View this message in context: http://sox.10957.n7.nabble.com/a-way-to-get-timestamps-of-silence-tp5157.html
Sent from the SoX mailing list archive at Nabble.com.

On Sep 26 10:42:01, jim...@gm... wrote:
> >From some PC's and androids we are getting WAV audio from the microphone with
> very short bursts of zeros periodically from an HTML5 browser in Chrome or
> Firefox.  I do not know why this is happening and presume it is a bug.

Can you point us to the page where these recordings originate?

> The zeros are screwing up processing of the audio downstream.
> I am looking for a way to mitigate this issue with sox so that
> the zero glitches do not break the downstream system.
> I am told that replacing the silence with noise is about the best I can do.

Can you please describe how the silence screws up your processing,
and how noise instead of silence makes it better?

> Currently I use a sox filter to resample the audio down to 8k and tried
> removing the zeros in sox but am not if I can do that
> or what the parameters should be.

Why are you downsampling to 8k first?
What is the original format of the audio,
and why is it not suitable as is?

> I tried
> sox - -t wav -b 16 - rate -I 8000 dither silence 1 0.2 0% -0.1 0.2 0%  -s
> That does not seem to remove all the short bursts of zeros.  I would like a
> suggestion on what parameter values I should use to remove the zero bursts.

Isn't the below-periods (-0.1) supposed to be an integer?
Is that a bug then that SoX accepts -0.1 as a value?

        if (sscanf(argv[0], "%d", &silence->stop_periods) != 1)


Also, there is no '-s' option to the silence effect.
It seems to be silently ignored (as is '-X -Y -Z' in that place).
Is that a bug?

Can you show us a (short) recording that exhibits this
(in the original format that comes from the browser?)


	Jan

On Sep 12 11:43:19, mar...@mu... wrote:
> Is there a way for the spectrogram effect to select
> a specific frequency range for the Y axis to be displayed,
> e.g. between 200Hz and 1000Hz?

No, the spectrogram effect itself cannot do it.

> If the sox effect does not implement this, (maybe this is a crazy idea)
> but could there be a workaround to first frequency shift the sound down
> by 200Hz and then downsample the audio to 800Hz (samplerate=1.6kHz)?

If you are willing to change the audio before doing the spectrogram,
it seems easier to just bandpass the frequency range you want.

	Jan

>From some PC's and androids we are getting WAV audio from the microphone with
very short bursts of zeros periodically from an HTML5 browser in Chrome or
Firefox.  I do not know why this is happening and presume it is a bug.  The
zeros are screwing up processing of the audio downstream.  I am looking for
a way to mitigate this issue with sox so that the zero glitches do not break
the downstream system.  I am told that replacing the silence with noise is
about the best I can do.

Currently I use a sox filter to resample the audio down to 8k and tried
removing the zeros in sox but am not if I can do that or what the parameters
should be.  I tried


sox - -t wav -b 16 - rate -I 8000 dither silence 1 0.2 0% -0.1 0.2 0%  -s

That does not seem to remove all the short bursts of zeros.  I would like a
suggestion on what parameter values I should use to remove the zero bursts.

But, I am told that a better solution would be to replace the zero bursts
with noise.  How can I accomplish that with sox?

Thanks,

Jim



--
View this message in context: http://sox.10957.n7.nabble.com/How-can-I-replace-absolute-silence-zeros-with-noise-in-HTML5-wav-audio-tp5154.html
Sent from the SoX mailing list archive at Nabble.com.

Is it possible to have sox create a ping-pong loop of a certain time
length  (when end is reached, go back to beginning in reverse)?

For those who don't know what a ping pong effect is

The sample is played forwards and backwards
in a loop between the starting and ending position; at the beginning,
it's played forwards.  When the end position is reached, it immediately
(with no pause at all) plays backwards until the starting position, and
when that is reached it plays forwards again until the ending, etc, etc,
ad infinitum.

Aloha
(sorry for the second post)

Is it possible to have sox create a ping-pong loop  (when end is reached,
go back to beginning in reverse)?

aloha

Hi Marlon,

as a workaround, one could use the frequency mixing, that occures due to
an amplitude modulation, to shift the spectrum. So you could:
1. cut out the relevant part of the spectrum with some bandpass filter(s)
2. multiply with a sine-signal with the lowest frequency, e.g. 200Hz
3. cut at the high frequency shifted by the low one (e.g. 800Hz [1]).

1] if the bandwidth you want to observe is larger than twice the lower
frequency, you first have to shift your signal upwards with a similar
trick, otherwise the upper and lower sidebands are mixed. Make sure to
increase sample rate if nescessary (step 1.25), to shift by a larger
amount as your bandwidth (step 1.5) and to cut out the relevant sideband
again afterwards (step 1.75).

Step 3 is nescessary and cannot be replaced by simple downsampling,
because else you would fold the higher frequencies into the range you
are looking at (e.g. if you sample down to 1.6kHz, a signal at 2kHz will
appear at 1.2kHz and 400Hz). See aliasing effect
(http://en.wikipedia.org/wiki/Aliasing).

Sox provides you with all the tools you need for this, look at
effects: synth, bandpass
combining method: multiply

Greetings,
Erich

P.S.: I like to help out with crazy ideas :-)

Am 12.09.2014 17:43, schrieb Schumacher Marlon:
> Hello list,
> 
> Is there a way for the spectrogram effect to select a specific frequency range for the Y axis to be displayed, e.g. between 200Hz and 1000Hz?
> 
> If the sox effect does not implement this, (maybe this is a crazy idea) but could there be a workaround to first frequency shift the sound down by 200Hz and then downsample the audio to 800Hz (samplerate=1.6kHz)?
> 
> Thanks,
> Marlon

Is it a coincidence that Marcus Bandiera just asked the same question?

Would it be simpler to just crop the spectrogram image once it's created? Then at least the scale values will be correct. You could use image magick or something to automate it if necessary.

Peter Shute

Sent from my iPad

> On 13 Sep 2014, at 1:44 am, "Schumacher Marlon" <mar...@mu...> wrote:
> 
> Hello list,
> 
> Is there a way for the spectrogram effect to select a specific frequency range for the Y axis to be displayed, e.g. between 200Hz and 1000Hz?
> 
> If the sox effect does not implement this, (maybe this is a crazy idea) but could there be a workaround to first frequency shift the sound down by 200Hz and then downsample the audio to 800Hz (samplerate=1.6kHz)?
> 
> Thanks,
> Marlon
> ------------------------------------------------------------------------------
> Want excitement?
> Manually upgrade your production database.
> When you want reliability, choose Perforce
> Perforce version control. Predictably reliable.
> http://pubads.g.doubleclick.net/gampad/clk?id=157508191&iu=/4140/ostg.clktrk
> _______________________________________________
> Sox-users mailing list
> Sox...@li...
> https://lists.sourceforge.net/lists/listinfo/sox-users

Hello list,

Is there a way for the spectrogram effect to select a specific frequency range for the Y axis to be displayed, e.g. between 200Hz and 1000Hz?

If the sox effect does not implement this, (maybe this is a crazy idea) but could there be a workaround to first frequency shift the sound down by 200Hz and then downsample the audio to 800Hz (samplerate=1.6kHz)?

Thanks,
Marlon

Can I select frequency spam (Y axis) in Spectrogram effect in order to
analyze a limited portion of audio band (i.e. form 1000 Hz to 2000 Hz)?
Thank you very much.

Marco Bandiera - Italy