1

Let's say, during a live performance I save the mic input of every instrument and vocals. I also record the speaker output from the crowd.

Can I extract the volumes of the individual instruments from the speaker/crowd audio? (Disregarding crowd noise)

Is there a way I can do this via audacity, scripting, or any software? I am only interested in the volume range of each instrument!

This is a computer science student project that hopes to automate sound checks for band performances at make-shift locations.

  • Just to clarify, it sounds to me like you multi-track recorded the band live, plus an ambience mic for the final mix + audience, and you want to figure out the volume of each given component (the multi-track recording) in the ambience recording. Is that right? – AaronD Dec 2 '16 at 6:43
1

Can you isolate the intended instrument or voice for each mic input?

NO. There is no practical way of doing this. This is the "holy grail" of audio processing that people have been pursuing for decades without success.

Can you predict the levels from different microphones in some attempt to "automate" the setup?

NO. Especially in transient setups (different venues, different acoustics, different physical configurations, etc. etc. etc.) There are millions of variations which make this an impossible task on a practical basis.

Can you use multi-track recordings to "simulate" performance as an aid for sound-system setup and initial level settings?

YES There are several simple multi-track recorders that can be plugged into the "effects loop" of a mixing desk. They can record ALL the channels of a performance, and then they can be used to play back ALL the channels in the same levels as the performance. That way the sound people can do a first-pass equalization and mix settings before the performers are avilable for a proper sound-check.

Some of the recorders that have been made specifically for this process include products from JoeCo, and Cymatic Audio.

1

It depends on the quality of the recording/venue/PA. The basic technique you use here is echo cancellation like that used in mobile phones: you try subtracting a filtered version of the original signal from the composite signal until they are no longer correlated.

You'll have much better likelihood for success if you record the total mix before going through the PA (basically, using an analog mixer but trying to recover the volume settings for the sake of starting the digital mix automation with it). Pop concert acoustics are a mud pit. There will be so much reverbation in it that recovery of correlated components will be very, very iffy.

In addition, acoustics are quite different with and without audience. So are noise levels.

0

A) Mixing music is not an automated process, it is a creative process driven by a combination of the mixers operational ability and their understanding of the music.

B) Every instrument will be able to play a wide range of notes and loudness, however the level they will actually play during performance will depend on the music requirements.

Having said that, you need to be researching the following topics:

  1. Music Interfaces
  2. Analogue to Digital Conversion
  3. Signal Processing
  4. Loudness management (R128/LUFS etc)
  5. Scripting and Coding
  6. Possibly Puredata, MaxMSP, Portaudio

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.