Combine Multiple Audio Files into a single higher-quality audio File

BACKGROUND

  • My team gave a demo to a large audience - we recorded the audio of the demo in multiple locations in the room (3)
  • the audio was recorded using cheap laptop microphones
  • I was not involved in the recording of the audio or the demo
  • Both audio files suck in some form
  • the first one is of a recording near the speaker - which clearly gets his voice but the the audience is muffled - also this one is slightly noisy
  • The second recording was done in the middle of the audience - it gets the audience questions clearly but actually gets the speaker rather sometimes well and sometimes poorly (not all the speakers spoke loudly enough to be heard)

MY QUESTION

  • Is there any techinque or software which can be used to merge these audio files in such a way that the best qualities of each are preserved.
  • I am NOT asking now to simply merge them together in one track - I've already done that in Audacity and it is certainly better - what I am looking for could be considered closer to how HDR images are created - multiple exposures combined into an enhanced new version which is not simply an average of the inputs.

NOTE

  • Am not an "Audio" guy - just a normal user

With Audacity, you will be able to achieve close to everything you want to achieve provided you understand how to work audio. It is important you need to be able to visualize sound in terms of identifying the different frequencies. Frequencies are like the audio equivalent of an image's histogram, and to achieve "HDR" audio, you will need to blend in multiple tracks that has emphasis on different frequencies.

What I recommend for you to do is this (assuming you got two sources of audio):

  1. Line up all the individual sources of audio - they should sync or be as close to syncing as possible from start to end.
  2. For each source of audio, in Audacity, split it up into four tracks each. You should now have eight tracks, 4 from each original audio stream.
  3. Clean up noise for only ONE track of each audio stream, and use the Equalizer function to best boost the important audio in this track (e.g the speaker).
  4. Use the Equalizer function, adjust the other 3 so the emphasis is on the lower band freq, middle band freq, and upper band freq.
  5. To recap, you should have 8 tracks in a single project in Audacity, 4 from each audio source, 1 of the 4 is denoised and important audio boosted, and the other 3 cover a range of lower freq, middle freq, and upper freq audio.
  6. On the left hand side of each track that's showing the waveform, you should see a volume adjuster - this comes the most time consuming part. Adjust the volume for each track individually as you play the the entire project over and over again. Fine-tune it to your desired results.
  7. Export as the audio format of your choice - you should now have a piece of HDR audio.

Again, imagine how a photo does HDR - by combining 3 or more photos (of the same shot) with shadow details, with highlight details, and also boosting mid-range tonal details. Similarly, with audio, you combine 3 or more audio tracks that covers the lower freq, middle freq, and upper freq, and have another track that covers critical audio.

Good luck - it will take some work, but I am confident with this method it will get you if not what you want, at least very close to it.


I don't know of one, but it is in theory possible. See the ‘Noise removal’ function in Audacity (and similar functionality in other audio editors); it might be possible to extend this sort of processing by correlating the signal:noise over multiple files.

Of course one problem with this would be that you'd have to sync the signals exactly (and note that audio recorded on two devices without a common clock will naturally drift off sync slowly over the course of a longer recording), and account for the different imaging (stereo, reverberation, frequency response) of the different inputs. Not necessarily that easy to write.

For now I'd probably just use the front source (run through Noise Removal, assuming you can get a silent section of it for noise profiling), and fade the other source in just when required to get the audience questions. You can do that in Audacity, but as its multi-track editing is pretty weak you might be better off with a dedicated multi-track sequencer such as REAPER.


sadly, no. I suppose you could combine the two as seperate tracks initially, have them sync up (audacity does this) and cut out segments as needed.


Your HDR-Audio reference is what also hit me in the first point itself.

  1. You would need to 'clean up' the individual audio files to retain the elements you are interested in at the highest amplitude and suppress other 'noise'.
    • Then you would need to 'mix' them (as is done in studio recordings) with suitable amplitude levels

I do not know of the right tools (or, if Audacity itself can achieve this),
But, would like to know if this can be done with freeware.