Originally posted by Dave2002
View Post
Indeed - I agree. However, if the specification of the synthesizer is known - and this should be particularly true with digital ones - do correct me if I'm wrong - if all the data is the same in the Midi data, then the effect "should" be well defined, and hence the sound should be too.
OK - I think you are going to correct me ....
If the system is deterministic, then the audio should be well defined, so if a sound engineer or composer generates some sounds and keeps the Midi data, then the audio produced (which could be checked by recording it as an audio file as it is generated) "should" be the same. However, if the delays and other factors in the sound generation are non deterministic - which they may be - due to delays etc, then even using exactly the same code and data could lead to different sounds. That is an area where you will have much more experience to date that I have, and perhaps you sometimes use Midi, and sometimes use audio files - and sometimes both - for that very reason.
My expectation would be that if one had access to a high quality synthesizer - either digital, analogue or virtual (hosted inside another system) it would be good to use that instead of possibly lower quality audio files, but otherwise it might be better to work with higher quality audio files, than a lower quality synthesizer working off "near equivalent" Midi data. As I wrote in the previous paragraph, sometimes it may be desirable to use both audio and midi representations, and also sometimes it might be useful to have lower quality audio files as well as high quality ones.
A lot will depend on what stage in the sound production these data components are required. Perhaps, in a similar way to video editing, sometimes lower quality audio files are used either for speed of production, or to minimise storage space, or to minimise data transfer requirements, or to provide a "prototype" version for auditioning, but at the production stage the highest available audio data will be used. Clearly live situations are more demanding in that respect than off-line sound development.
OK - I think you are going to correct me ....
If the system is deterministic, then the audio should be well defined, so if a sound engineer or composer generates some sounds and keeps the Midi data, then the audio produced (which could be checked by recording it as an audio file as it is generated) "should" be the same. However, if the delays and other factors in the sound generation are non deterministic - which they may be - due to delays etc, then even using exactly the same code and data could lead to different sounds. That is an area where you will have much more experience to date that I have, and perhaps you sometimes use Midi, and sometimes use audio files - and sometimes both - for that very reason.
My expectation would be that if one had access to a high quality synthesizer - either digital, analogue or virtual (hosted inside another system) it would be good to use that instead of possibly lower quality audio files, but otherwise it might be better to work with higher quality audio files, than a lower quality synthesizer working off "near equivalent" Midi data. As I wrote in the previous paragraph, sometimes it may be desirable to use both audio and midi representations, and also sometimes it might be useful to have lower quality audio files as well as high quality ones.
A lot will depend on what stage in the sound production these data components are required. Perhaps, in a similar way to video editing, sometimes lower quality audio files are used either for speed of production, or to minimise storage space, or to minimise data transfer requirements, or to provide a "prototype" version for auditioning, but at the production stage the highest available audio data will be used. Clearly live situations are more demanding in that respect than off-line sound development.
A couple of things
1: MIDI is only a way of one thing controlling another. It doesn't make a sound, in some ways it's got very little to do with sound at all
2: It was a bit of a happy accident that the Atari 1040 St had MIDI ports fitted which kickstarted all sorts of experiments
3: You are assuming that the "goal" is to be predictable ... what Gershwin said
4: You are also assuming that music is "note-based" and synthesis always aims to create versions of acoustic sounds OR emulations of electronic instruments.
5: Two of the dullest things I can do with my computer are a: Fill in my tax return and b: Try to make it sound like a piano !
I've just done a 30 minute online gig.... I was playing (in Audiomulch)
1: Some recordings of bass clarinet improvisations (which I manipulated live )
2: A recording I made several years ago of the gamelan at the SBC which I played with live
3: The sound of bees (not live this time but sometimes I do stick microphones under a beehive)
4: A live stream of a South Indian singer who was improvising with what she could hear coming back (with a 14 second latency delay)
I was controlling the volumes and some of the filters and other parameters using a MIDI controller (knob box)
I wasn't (today anyway) playing any "notes" ... last time I also played the Flugelhorn live into the system but not today
This is fairly typical of how many folks use technology in performance
Comment