The overall goal when mixing a recording from a Sofar Sounds event is to to get an honest representation of the performance. This audio will be used for a video production, so keep it as real as possible. If it sounds too studio-like it won’t appear to be honest.
We make two separate outputs of audio:
- Artist approval – just the song to be loaded into Soundcloud for the artist to listen to
- Video sound – some introduction/pre-amble and audience clapping at the end
What the video people are looking for here is plenty of scope to work with. By that I mean leave plenty of audience reaction (clapping) and plenty of pre-amble (the talking before the song) in the final output. When it comes to video editing, it’s much simpler for them to use one continuous stream of audio and put the right visuals around it. They are very capable of fading in and out at the right start points. It’s best to leave it up to the video editor to choose the appropriate start and end points (we don’t need to assume this).
This performance by New Jungle Order is a good example what of what was needed to make the video complete.
Video Output Format
- File: 16 bit, 48 kHz uncompressed – (some) video people are very fussy about getting 48 kHz files. It’s best not to compress it at this stage as this video editing software will do so anyway. Compressing it multiple times will just make it sound worse anyway.
- Loudness: -14 LUFS – this is a rule of thumb, but, at this level Youtube will be happy.
- Limiting: -1 dB – make sure there are no peaks beyond -1 dB. This will reduce the chances of inter-sample distortion when being played back.
- Dithering: definitely – don’t ever forget to dither when creating 16 bit files from higher resolution material.
Artist Approval Sound
I use the same mix as I’ve done for video – I take the video output track and then do the following:
- Trim off any leading talk (pre-amble). I make the first note of the song the very beginning of the track.
- Fade out during the audience clapping – Gradually fade from half-way in the audience clapping until the end of the clapping.
- Normalise to -1 dB – This will be compressed so make sure don’t create any chances of inter-sample distortion.
- Resample to 44.1 kHz – This is just to be more friendly with the likes of Soundcloud.
- Dither to 16 bit – Just to get the best sounding 16 bit file.
- Compress at 320 kbps AAC – AAC can be decoded by practically everything. Going for the maximum will yield the best results when uploading to Soundcloud.
I typically leave the audience mics playing through the whole track. By keeping some of the sound of the room in the final output, it makes the video sound a little more authentic. Specific things I do are:
- Automate the audience bus to drop around 12 dB during the actual song.
- Add a compressor to the audience bus which is side-chained from the vocal bus. Use this to take around 6 dB off the audience bus when people are talking in the vocal mics.
- It doesn’t hurt to make the audience mics a little brighter than what you think at first.
The most common configuration I use for audience mics is a pair of ribbon microphones in a mid-side configuration. To process this, I do the following:
- Route the “mid” track to the audience bus with a high-shelf EQ that boosts 9 dB (ish) around 6 kHz.
- Route the “side” track to the audience bus as well, plus:
- Use a high-shelf EQ to boost about 12-15 dB around 8 kHz
- Use a gently (6 dB / octave) high-pass filter around 160 Hz
- Use a utility plugin to convert this track to stereo
- Use a utility plugin to flip the phase on the right channel only
- Use an EQ on the audience bus to reduce the level of the “nodes” in the room
- Use a compressor to add a little more body to the audience sounds
Have a listen for phasing problems. This can be rather difficult in live scenarios like this. Sometimes it is better to mute a mic and work around the problem.