[JEK3]

Hi,
 
I'm sometimes reading this forum and I always found it very interesting. I saw that many people couldn't find a decent pink noise source so I decided to create something and to join this forum. I just uploaded some Pink noise and some White noise on my youtube channel. While I myself consider youtube audio quality too low, I thought it could anyway help some of you. Here are the links:
 
Pink noise: http://www.youtube.com/watch?v=6ZEFb6Zi3sI
White noise: http://www.youtube.com/watch?v=iEnRj1h9Deo
 
The audio files have been created at 24bit/96kHz, then downgraded to the best possible YT quality.
 
I'm planning to create more audio files. I'm thinking about some sweeps and eventually some test tones. Please share your suggestions.
 
I'm also considering to create a CD (and eventually release the 24bit/96kHz files too). I'm thinking about something like 10 min white noise - 2 min. pause - 10 min pink noise - 2 min pause - 10 min 20Hz-20kHz sweeps - 2 min pause - and then 1 min x frequency - 1 min pause for various frequencies (probably around 10 frequencies, like 50Hz - 120Hz - 250Hz - 600Hz - 1200Hz - 2000Hz - 4000Hz - 6000Hz - 8000Hz - 12000Hz - 18000Hz). Nothing has been defined yet, so please let me know if you would find that useful, what you think about it and share your ideas!

 

[stv014]

Quote:

Hi,
 
I'm sometimes reading this forum and I always found it very interesting. I saw that many people couldn't find a decent pink noise source so I decided to create something and to join this forum. I just uploaded some Pink noise and some White noise on my youtube channel. While I myself consider youtube audio quality too low, I thought it could anyway help some of you. Here are the links:
 
Pink noise: http://www.youtube.com/watch?v=6ZEFb6Zi3sI
White noise: http://www.youtube.com/watch?v=iEnRj1h9Deo
 
The audio files have been created at 24bit/96kHz, then downgraded to the best possible YT quality.
 
I'm planning to create more audio files. I'm thinking about some sweeps and eventually some test tones. Please share your suggestions.
 
I'm also considering to create a CD (and eventually release the 24bit/96kHz files too). I'm thinking about something like 10 min white noise - 2 min. pause - 10 min pink noise - 2 min pause - 10 min 20Hz-20kHz sweeps - 2 min pause - and then 1 min x frequency - 1 min pause for various frequencies (probably around 10 frequencies, like 50Hz - 120Hz - 250Hz - 600Hz - 1200Hz - 2000Hz - 4000Hz - 6000Hz - 8000Hz - 12000Hz - 18000Hz). Nothing has been defined yet, so please let me know if you would find that useful, what you think about it and share your ideas!

 
Anyone can create those tones with my testgen utility and the following simple text file that defines the signals to be generated:

 noise 0 600 0 0.4 0.4 noise 720 600 -0.5 0.4 0.4 pulse 1440 s1 t2 0 -1 -1 sine 1440 600 20 1 20000 1 0 d-3 d-3 sine 2160 60 50 0 d-3 d-3 sine 2280 60 120 0 d-3 d-3 sine 2400 60 250 0 d-3 d-3 sine 2520 60 600 0 d-3 d-3 sine 2640 60 1200 0 d-3 d-3 sine 2760 60 2000 0 d-3 d-3 sine 2880 60 4000 0 d-3 d-3 sine 3000 60 6000 0 d-3 d-3 sine 3120 60 8000 0 d-3 d-3 sine 3240 60 12000 0 d-3 d-3 sine 3360 60 18000 0 d-3 d-3

Much smaller download, and the user can choose the sample rate and resolution.

 

[JEK3]

Very nice, thank you!!!

 

[stv014]

The pink noise has perhaps too much inaudible very low frequency "rumble", though, because the 3 dB/octave slope goes lower than necessary; I may filter that out in the next version.

 

[Prowler]

I made a Winamp plugin a while ago to generate pink noise... I learned it's not so trivial how to generate pink noise (compared to white or brown).
http://www.winamp.com/plugin/the-colours-of-noise/221540

 

[stv014]

A simple method is to sum the output of white noise generators that are upsampled (using simple sample duplication) by powers of two. That is, one normal white noise generator, then one that repeats each sample twice, then another one that repeats each sample four times, and so on (each additional generator reduces the lowest frequency the 3 dB/octave slope extends to by one octave). This algorithm is cheap to compute (only 2 random numbers are needed on average by output sample, and log2(N) sample and hold and summing operations for a lowest frequency of Fs/N), but there is a slight (within 1 dB) deviation from a perfect 3 dB/octave response.
 
The other approach is to filter white noise with an IIR or FIR filter. The former is not easy, because IIR filters normally roll off by multiples of 6 dB/octave, so it requires a relatively complex filter to approximate pink noise well. I used the FIR method, as it is fast enough on current computers, and it allows for creating a response that is both very accurate, and easily configurable.

 

[audiosampling]

Quote:

The audio files have been created at 24bit/96kHz, then downgraded to the best possible YT quality.
 

 
http://wavtones.com generates downloadable noise files on the fly - without any degradation - offering an extremely wide variety of noises (white, pink, brown, blue, impulse, ...) and sample rates up to 96kHz. It even produces filtered noises (with different filter slopes) and filtered noise sweeps (a great alternative to sine sweeps).