Is it ultimately better to resample a 22KHz sample to 44.1KHz using Renoise’s wave renderer (outputting as 44.1KHz 16-bit stereo will of course resample all lower quality samples to this), or using Sound Forge?
In Sound Forge I can apply an anti-alias filter while resampling, but I’m not even sure if I need this or not. The thing is that I have a 22KHz mono wave file (I still don’t get this, as I thought a 44.1KHz stereo wave has a 22KHz wave per channel) that I want to parametrically equalise using a VST plugin. When the input is a 44.1KHz wave, the plugin window shows a graph from 0-22049 Hz, allowing you to enter values in that frequency range. However, when I input that 22KHz sample, the range of the graph is only 0-11024.
When I set the sample rate to 44.1KHz only (without resampling) in Sound Forge, and use a peak EQ at 15000Hz, I still get an audible boost. So I’m not really sure what is going on here. I do want to have a boost at 15000Hz, however, with a Q of 1.0 and a gain of 0.4dB. So is it safe to just set the sample rate, and equalise the sample as if it contained those frequencies? (Of course when changing the sample rate, the sample becomes twice as fast, but I then play the note back an octave lower in Renoise - not sure if this preserves maximum quality, though.)
And can someone please enlighten me why some mono samples are 44.1KHz? Isn’t 22KHz all you need if it is mono? (provided you are working at CD-quality)
No. It has 44.1kHz sampling rate per channel. Sampling rate is independant from the number of channels.
I think you’re mixing up sampling rate with the frequency content of the sample. Humans hear up to around 20kHz, and to produce a digital recording which contains frequencies this high, we have to sample at twice this rate due to the Nyquist theorem.
Check this link for a more complete explaination.
As for your dilemma, I would resample your sample to 44.1kHz before importing it into renoise. Not that I think renoise doesn’t do a good job, but just to keep things simple by having all your samples at the same quality.
this is because if you have a signal (sound) sampled at Xkhz frequency, you can only reproduce the specturm of that signal ranging from 0 to X/2.
This is called the Sampling Theorem
probably because a Gaussian filter is used there, meaning that the frequencies affected by the boost are a range in which 15khz is just the middle point. The width of this range is determined by the Q parameter you mentioned.
I suggest you to read more about audio signals editing : there are tons of good theory sites around, just search on Google for some of the words written with Italic characters in this post (apart from these last two…
Thanks for all the help. I think I’m beginning to understand this a lot more. And kmkrbes, I do think I was mixing the two up. So this 22KHz sample I have is theoretically only able to contain frequencies from 0-11024Hz, as the EQ plugin graph showed. And It-Alien, I think you’re right about it using a filter of some sort, it was just that I was doing it the wrong way first and it didn’t actually tell me it was using an interpolation or something.
So the final verdict would be to resample all my samples to 44.1Khz in Sound Forge before loading them into my song in Renoise? (as I noticed there were a few others at 22KHz too)
And if sampling rate is independent from the number of channels, why all the 96000KHz and 192000KHz sampling rates you hear about as used in studios and 5.1 surround setups? So with 96000KHz you can reproduce frequencies from 0-48000Hz? (the point being?) And 192000KHz? Can’t remember if this 100% accurate, but I may have heard/read about such rediculously high numbers somewhere…
DVD audio uses a 96kHz sampling rate. I’ve no idea what supports 192kHz, other than editing suites.
The argument is that we can indeed hear frequencies over 20kHz, not directly, but through their modulation effects on lower, audible frequncies. Sampling at 96kHz or higher lets us capture these ultrasonic overtones.
Some people believe it (it does make sense – the sounds you hear in the world are not limited to any sampling rate), others don’t (and think it’s just the industry making sure everyone needs to upgrade their equipment once again)
But the fact remains: a majority of the populace don’t know or care what sampling rate is and just want to hear some good music.
I have a pretty new soundcard which supports 192kHz. It may be false or not that the human ear can hear the difference over 44.1kHz, but some claim that there really is a big difference. In either way it can’t hurt to go up in sampling and bit rate. Now when HD and RAM is cheap it makes sense.
Anyways… if you work with 96kHz samples and apply different FX, you will be able to modify it much more before it looses quality compared to 44.1kHz. I guess this applies even more to 32 bit compared to 16 bit. Especially for post processing as Renoise uses high quality internally. Render to as high quality as possible and downsample to CD quality or MP3 as a last step and then you are sure that the least possible quality was lost at the way.
and sampling in 96kHz has another big advantage : if you play the sampled tone one octave higher you will hear only the same artefacts as if you’ve sampled it with 44kh in this upper octave.
i think another good thing about 96khz is when sampling hi-freq content from analog source (i.e. cymbals), possible quantisation noise (which is around nyquist frequency) will be above audible range. so after sampling you can simpy filter everything above 22.05khz, then resample to 44.1khz and have noise-free sample…
i imagine only very expensive soundcards have analog filters builed in to filter the signal before digitizing?
