Audio processing for GUADEC videos
Open the file in Audacity
Depending on your installed version of Audacity and its build options, you may be able to open the video file to automatically import the audio.
If that does not work, you can use a tool to export the audio from the video file:
Transmageddon is a GUI tool: choose No container as output format, FLAC as audio codec, No Video as video codec. The resulting file should be saved in your ~/Music directory.
ffmpeg is a command line tool: ffmpeg -i input.mp4 -vn output.wav.
Improve the sound
Not all of these improvements will be necessary and you may want to apply them in a different order. Use your ears and best judgement to find out what works, and try to not overdo it.
Noise reduction is used to remove background noise such as hisses.
- Zoom in and find a place where there is only background noise (the longer the better, one or two seconds should be good).
- Go to Effect → Noise removal, click the "Get Noise Profile" button in the Step 1 frame.
- Select the whole file (ctrl + A)
- Go to Effect → Noise removal and adjust values in the Step 2 frame and hit the Preview button to check what this would sound like.
You do not want to go too strong or the resulting audio will sound weird (metallic/bubbly artefacts). 16dB for Noise reduction and 0.1 for Attack/decay time seem to work fine, but you may need to be even more subtle.
Dynamic range is the difference between the lowest and highest sound volumes. While dynamic range is a generally a good thing as it adds nuance, you may find that sometime the speaker doesn't talk loud enough, or is too far from the microphone, or someone in the audience doesn't have the mic and asks question. Also there could be some peaks due to unwanted noise sources (someone coughing, something accidentally dropped on the table).
You can try reducing dynamic range with Effect → Compressor.
You can use Effect → Equalization to boost or cut some frequencies.
I'm not sure yet what's best, maybe someone with better knowledge of this topic could help improve this area. Is the RIAA preset (bass boost, treble cut) a good idea?
The sound volume is not very high, mostly due to the cameras being far from the speaker and not having an external microphone. You can use Effect → Amplify to fix that. Don't allow clipping or it will saturate. If a specific part of the talk is at a lower volume, you can select it and apply the effect to that part.
Putting it back together
In audacity choose File → Export , give it a name such as improved, leaving WAV as output format.
You can create a new video file with the new audio using ffmpeg -i input.ogv -i improved.wav -vcodec copy -map 0:v:0 -map 1 -ab 128k improved.ogv