Sign in
Search audiodsp
Subscribe to audiodsp
audiodsp by Keywords
Discussion Groups
Discussion Groups | Audio Signal Processing | RE: Speech Noise
Technical discussions related to Audio Signal Processing (digital effects, acoustics, noise reduction, musical signal processing, etc).
Speech Noise - marzaq7 - Mar 6 6:55:00 2002
Hello all, I need to generate speech noise. I know that speech noise is obtained by filtering white noise above 1000Hz at the rate of about 12dB per octave. My question is what type of digital filter and coefficients do I need to use to the white noise above 1000Hz fall at a rate of 12dB per octave ? Thanks !
(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )
Re: Speech Noise - Marcio Zaquela - Mar 7 12:40:00 2002
Hi, thank you for your response. The goal is to design an Audiometer that generates a sound noise called "speech noise". It means to generate white noise (random numbers) and filter it above 1000Hz at the rate of about 12dB per octave. As you said, I need a Low Pass Filter, but the hard part is finding the correct "slope" to get the 12db per octave. Any guess ? Regards, Marcio ----- Original Message ----- > On Wed, 06 March 2002, "marzaq7" wrote: > > > > > Hello all, > > > > I need to generate speech noise. I know that speech noise is obtained > > by filtering white noise above 1000Hz at the rate of about 12dB per > > octave. > > > > My question is what type of digital filter and coefficients do I need > > to use to the white noise above 1000Hz fall at a rate of 12dB per > > octave ? > > > > Thanks ! > > > > Speech Noise? I can only assume you mean "White Noise in the Speech Spectrum". The normal range of speech is from 200hz to around 2000hz. A quick test with a spectrum analyzer will confirm this. > You will need a Low Pass Filter. The hard part is finding the correct "slope" to get the 12db per octive. 12 db per octive will give it a very steep slope. You won't have much signal left after a 12 db reduction. I have a graphic eq if you need one to play around with. By the way, what is the goal here? There maybe other avenues of approch to this problem. > > ET >
(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )
RE: Speech Noise - Egler, Mark - Mar 7 15:20:00 2002
Marcio,
If you design
a second-order low-pass Butterworth "analog-prototype" IIR filter, which just about any digital
filter design package can do, you should get 12 dB/octave slope since that is what a true
analog 2nd-prder Butterworth will do. The digital IIR approximation of it will get
steeper nearer the Nyquist frequency (sample-rate/2) due to the frequency-warping effect of the
IIR design approach, but if your sample rate is much higher than the cut-off you should be
OK.
Mark
Egler
-----Original Message-----Hi, thank you for your response.
From: Marcio Zaquela [mailto:m...@terra.com.br]
Sent: Thu, March 07, 2002 11:41 AM
To: e...@woodstock.com; a...@yahoogroups.com
Subject: Re: [audiodsp] Speech Noise
The goal is to design an Audiometer that generates a sound noise called
"speech noise". It means to generate white noise (random numbers) and filter
it above 1000Hz at the rate of about 12dB per octave.
As you said, I need a Low Pass Filter, but the hard part is finding the
correct "slope" to get the 12db per octave.
Any guess ?
Regards,
Marcio
----- Original Message -----
> On Wed, 06 March 2002, "marzaq7" wrote:
>
> >
> > Hello all,
> >
> > I need to generate speech noise. I know that speech noise is obtained
> > by filtering white noise above 1000Hz at the rate of about 12dB per
> > octave.
> >
> > My question is what type of digital filter and coefficients do I need
> > to use to the white noise above 1000Hz fall at a rate of 12dB per
> > octave ?
> >
> > Thanks !
> >
>
> Speech Noise? I can only assume you mean "White Noise in the Speech
Spectrum". The normal range of speech is from 200hz to around 2000hz. A
quick test with a spectrum analyzer will confirm this.
> You will need a Low Pass Filter. The hard part is finding the correct
"slope" to get the 12db per octive. 12 db per octive will give it a very
steep slope. You won't have much signal left after a 12 db reduction. I have
a graphic eq if you need one to play around with. By the way, what is the
goal here? There maybe other avenues of approch to this problem.
>
> ET
_____________________________________
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )
RE: Speech Noise - Ricardo Pucher - Mar 7 23:17:00 2002
That wouldn't be a problem, considering you could use a Second Order Butterworth filter. If you like to have two poles instead at 1000 Hz you should try to discretizice a suitable two-pole function. Remeber each pole adds -6dB per Octave. Am I correct? Lucas Pucher. > Hi, thank you for your response. > > The goal is to design an Audiometer that generates a sound noise called > "speech noise". It means to generate white noise (random numbers) and filter > it above 1000Hz at the rate of about 12dB per octave. > > As you said, I need a Low Pass Filter, but the hard part is finding the > correct "slope" to get the 12db per octave. > > Any guess ? > > Regards, > Marcio >
(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )