On 26/06/05, Goran Finnberg wrote:
> Eric Jacobs:
>> Impulse noise has much higher frequency harmonics than the audio
>> signal - these harmonics are not audible. However, DSP algorithms
>> can more clearly distinguish between noise and signal when these
>> higher frequency harmonics are present, and then more cleanly
>> remove the noise without affecting the signal.
> I must point out that there is a lot more to click detection and
> removal in the software than losing the slight frequency content above
> 20 kHz.
> I'm thinking how the software uses a combination of time based and
> amplitude (energy) based factors that are used in decide what is and
> is not an objectionable artifact.
> BTW, if a particular software uses only amplitude detection for clicks
> then I entirely agree that 192 kHz is a must together with >70 kHz
> bandwith, the best I've seen, in cartridges.
You can't use amplitude. You would use rate of change of amplitude,
compared with the average rate of change for that section of the sample.
>> and then more cleanly
>>> remove the noise without affecting the signal.
> Here I don't agree at all.
> If something is flagged as a click then a good algorithm will remove
> it with no audible artifact.
Yes. It is the detection that is helped by a higher sample rate.
[log in to unmask]