Sound Engineering: Why's it so hard to get things like timbre right?

Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
Sign up Log in
8 replies · 1K views
ardnog
Messages
52
Reaction score
29
Apologies if wrong forum. Could not see anything about acoustics. Is this mathematics?

Why's it so hard to get things like timbre right? Digitised instruments and voice changers don't always sound right.

I had a chance to visit a recording studio. Was given a lengthly talk about sound from a sound engineers point of view, and things like timbre, that we don't can't do properly yet on a computer, so that they can't make a proper male<-->female voice changer, or plugin that makes an electric instrument sound like an acoustic etc. Though, they can get close for certain instruments to the point where the majority of professionals are fooled.
 

Discussion

Physics news on Phys.org
To my mind one must be careful so separate the questions of recording (and reproducing) sound from that of synthesis of sound (like the voice changer).
In principal, given sufficient bandwidth, there is no reason the that digital sound reproduction cannot exhibit arbitrary fidelity. (Gnashing of teeth from analog aficionados). The synthesis part is much more problematic and requires a very large set of actively changing parameters. For instance, try to synthesize Miles Davis.
 
ardnog said:
Summary:: Why's it so hard to get things like timbre right? Digitised instruments and voice changers don't always sound right.

Apologies if wrong forum. Could not see anything about acoustics. Is this mathematics?

Why's it so hard to get things like timbre right? Digitised instruments and voice changers don't always sound right.

I had a chance to visit a recording studio. Was given a lengthly talk about sound from a sound engineers point of view, and things like timbre, that we don't can't do properly yet on a computer, so that they can't make a proper male<-->female voice changer, or plugin that makes an electric instrument sound like an acoustic etc. Though, they can get close for certain instruments to the point where the majority of professionals are fooled.
Timbre is a qualitative measure of the entire emitted audio signal; and even worse (from a modeling perspective) is that timbre can change over time (for example, the sound of a cymbal) because the relationships between individual frequencies that comprise the audio signal are not fixed in amplitude or phase as time progresses.

White noise is perfectly random- that can be simulated to some extent.
 
I suspect she was using the sound engineers definintion of timbre. It was explained to me as what makes a piano and guitar sound different even though they're playing the same note, and the same for a male and female voice. Likely there are many different things that contribute that make it hard to synthesise, but I'm surprised the machine learning people haven't managed it either.
 
The MP3 standard throws away some (or most) of the information, so we would not expect the reproduced sound wave to look the same as the original. And to trained ears, why should it not sound different?