Hi,
I am making an Earthquake sound.
It is about a 35Hz Noise with lots of subtle stuff done to it.
All goes well until Nyquist decides to save memory…
It, of spontaneity, reduces the sound-srate on my sound.
Although I can Force-Srate it back to my chosen rate… It will take charge of the situation and change it back to a slower rate.
Why this is a problem is Quantization Noise… That annoying wining sound heard in a sub-audible signal.
It manifests itself in a modulated tone at the sampling frequency…
Modulated by the first derivative of the slow-moving signal.
Even if I get out of a Self Written Nyquist Plugin at the chosen Srate (usually 44100),
Audacity upon manipulating the sound will intervene and drop the state:
So I ask two questions:
1) How do I stop Nyquist from taking ownership of MY Memory; AKA let me fix sample rates?
2) How do I stop Audacity from doing the same thing?
Even when exporting a sound in the MP3 format the quantization noise is still present.
I don’t know and I’ve never programed Nyquist but it’s probably not quantization noise…
And I have no idea how the sample rate can be accidently changed… That seems “impossible” unless you are copying someone else’s code and you don’t know what it’s doing. As you probably know, the sample rate isn’t part of the raw audio data. It’s a value “known” to Audacity (or player software, etc.). The sample rate is normally saved in the file header and if you edit a WAV header you can change the playback speed.
Quantization noise is related to bit depth (not sample rate). You can hear it at 8-bits, but you can’t hear it (under normal listening conditions) at 16-bit’s or better. Quantization noise is like a “fuzz” that rides on top of the signal. It’s similar to analog noise except it goes-away with digital silence. If you want to hear it, you can save any file as 8-bit WAV. (Like regular analog noise it will be more noticeable with quiet sounds.)
It might be aliasing which is “false frequencies”. That happens when the signal frequency is more than half the sample rate (the Nyquist limit). The higher frequencies are “folded back” below the Nyquist limit.
But aliasing doesn’t seem likely either with a signal around 35Hz. Even if the sample rate is reduced it’s probably not getting reduced that much…
So I’d guess your modulation or other processing is doing something unexpected.
Perhaps I am misusing the phrase “Quantization Noise”
I felt that that describes three different sundry effects:
The lack of resolution as you described in the number of analog quantization levels such as bits.
The effects of Nyquist folding attributed to a signal approaching half of the sample rate.
The time quantization effects of making a continuous signal into discrete temporal steps.
I am talking about effect #3 Nyquist has a property that if the signal is slow moving, it automatically reduces the sample rate to conserve mem.
That is a very good thing when memory is a concern. but I want a sound without the noise of TIME quantization of my very slow moving signal.
That is what I am describing with the Whine of the steps of a rapidly moving signal that is heard when the sample rate is in the audible region.
By forcing the sample rate back to 44100 the Whine goes away, but alas, Nyquist sees that I am wasting space and restores my sample rate to what they call the control-srate a much slower sample rate used primarily for envelopes and modulator functions.
I want to be in charge of how Nyquist handles my sounds.
This is perhaps a part of the warp header stuff that belongs to a sound, something that tells Nyquist NOT to mess with the sample rate.
It is just that I cannot find anything in the documentation as to limit the automatic feature of both Nyquist and Audacity in this regard.
Thanks for your input DVDdoug, it helped me to clarify exactly what I want.
P.S. I don’t know the technical terminology for type 3 (detailed above) Quantization Noise, If you know the name of this distortion, please do tell.
Thanks,
Dalagar
In Nyquist programming, there are two default sample rates:
*sound-srate*
*control-srate*
sound-srate is the sample rate of the selected track. control-srate is 1/20th of sound-srate.
control-sratemay be used by the developer to conserve memory for slow moving signals (such as amplitude envelopes or other “control signals”).
Nyquist will only resample a sound if your code tells it to.
“Quantization noise” occurs when the sample format is reduced. For example, reducing the “bits per sample” from 32 to 16.
The “sample rate” is the “number of samples per second”, and is unrelated to quantization noise,