This content has been marked as final. Show 9 replies
48 KHz stereo is the standard for DVD.
Thanks, Harm. At the risk of being obstuse I am asking why 48 Khz instead of the 44.1 KHz standard for CD's? If humans can't hear it, what does it add to the DVD? and the bit depth also puzzles me.Higher bit depth at lower sample rate makes better sound?
Maybe this is why I'm not a scientist/engineer!
I'm not positive about this, but I think they used 48KHz simply because most professional audio/video gear uses that same sample rate. This makes it easier to tramsfer audio from DV, DAT, etc. without the need for resampling.
As for why 48 in the first place? This has to do with the Nyquist limit... and that's way too deep for this thread!
Higher bit depth is another thing all together: This represents the number of possible volume levels (dynamic range) of each sample. 16 bit has 65535 levels which is a dynamic range of 96 dB. This is fine for final delivery -- assuming that the audio was well-mastered at an appropriate volume level.
During recording and editing, however, where it is likely that adjustments to dynamics (compression, expansion, noise gating, fading) and effects will be applied, it is best to use a higher bit depth so that less information is lost (quantized) on successive transformations of the audio. It also will keep broadband noise from creeping in when the audio is amplified.
For example, if you have audio recorded in 16 bit and then you amplify it to be 20 dB louder, your dynamic range has decreased to 76 dB and the noise floor has been raised -- meaning there is a constant whitenoise "hiss" in the background.
32 bit audio is different. The dynamic range is sufficient enough so that moderately large amplification will not neccessarily result in audible background noise.
The 32 bit audio is then downsampled during mastering to 16 bit for final delivery (often with aid dithering, noise shaping, etc. to avoid distortion and give the "illusion" of greater dynamic range)
I do agree that 48 KHz is a little bit silly for DVD when the audio is AC3 compressed anyway. I am fairly certain that none of these ultrasonic frequencies will survive AC3 encoding -- even at high rates.
The sampling rate (e.g. 44.1 or 48 kHz) determines the frequency range that can be reproduced. True enough, the above 2 rates do not have many difference. But 48 kHz was selected as the standard for digital video and DVD even when the CD rate of 44.1 kHz was omnipresent. It may be a bit easier for audio processing but I don't really know.
As for bit depth (12, 16 or 32) this determines the dynamic range: the amplitude is more precisely defined the higher the bit depth. This is necessary if you intend to process the audio with filters or other conversion. This is why Audition and Soundbooth translate up all audio to 32 bit depth. Premiere used to do it also (this was called conforming) but it was dropped for the later versions on account of user complaints that it was computer intensive.
With audio, 44.1k and 48k do not represent the range of hearing. These numbers indicate the number of samples take per second of an audio file. So 44,100 samples per second in the case 44.1k and 48,000 samples per second for 48k.
DVD spec is 16bit 48k.
>Premiere used to do it also (this was called conforming) but it was dropped
It was actually only dropped for DV media in a DV preset. Bringing in a .wav or .mp3 or non DV media, it will still be conformed.
According to Mr Nyquist, half the sample rate is the maximum frequency reproduced. So for 48k sampling you can record up to 24k audio frequency. Not many people can hear that high a frequency.
See http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem if you like maths.
Finally I can peck at you for something :)
>For example, if you have audio recorded in 16 bit and then you amplify it to be 20 dB louder, your dynamic range has decreased to 76 dB and the noise floor has been raised -- meaning there is a constant whitenoise "hiss" in the background.
NO, no, and no (except for that you'll have amplified the noise also)!
From start, you'll have a noise-floor and a maximum input level. If that has a 96dB dynamic range, the final dynamic range of your recorded sound will only depend on how close to the max input level you record.
If you record a steady sound 20 dB below max input level, your dynamic range of the recorded sound will always be 96db - 20db = 76dB, no matter if you amplify or not, it all lays in the recording stage.
Another thing is that a dynamic range at 76 dB is pretty good. It's counted as when the noise is 60dB below the signal, it's very hard to hear the noise. That's of course dependent of what kind of sound, if the sound itself has a big dynamic range, the noise level and the hearability of it will vary, then the magnitude of the noise-level, i.e. where the volume knob is set, determine how much noise you hear.
If it's white or pink noise in the "bottom" is sometimes hard to determine...
The 48kHz sampling frequency gives a bit more room than 44.1kHz for making good anti-aliasing filters (see David J's link), so that the max freq. you can hear (20kHz as a young stud) will be easier to present with for example less phase-shift compared to the original analog signal.
Dag, hit why DVD's went to 48k and Hi def formats are allowing 96k audio files, anti-aliasing.
David you are right, the sample rate to hearing rate is Sample rate/2, but if a frequency is above the range allowed by the sample rate it will wrap itself around and back into the hearable range as noise. Also, in many cases the higher frequencies get less samples than lower.