PodcastMastering.md

Editing

After recording, each person drops their .flac file into a sub-directory called Dropbox/UbuntuPodcast/{S09E00}, where S09 is the season number, and E00 is the number of this episode.

• Take a copy of the episode directory. Don't work inside the Dropbox directory because it annoys people when they get all your changes sync'd.
• Use the ~/Dropbox/UbuntuPodcast/Audacity/audacity.sh script to launch audacity (it kills and re-enables pulseaudio for you)
• Create a new project
• Import all the FLAC files as seperate stereo tracks. They will be slightly different lengths and volumes, but should all start at the same point in time.

Mix down

Mix down to mono - For each track:

• Select the whole track (Click the area below the L-R slider)
• Tracks -> Stereo Track to Mono
• Save

Align the tracks

Make sure all the presenter tracks are aligned to the "buscuits".

Top and tail

Delete everything before the start of Samantha's audio, and everything after the end.

• Click at the end of the head and use arrow down and left to select all tracks and back to the start.

Noise removal

Noise removal, known as Noise reduction on newer versions of Audacity.

• For each presenter track (not Samantha):
  • Select the longest section of the track with no talking
  • Effect -> Noise removal click Get Noise Profile
  • Select the whole track (Click the area below the L-R slider)
  • Effect -> Noise removal, leave the default settings and click OK
  • Save

Gate (optional)

If someone still has a lot of noise on their track, you can reduce it to silence when they aren't talking with the Gate plugin. This won't remove the noise completely, but will mean it won't interfere with other people speaking.

sudo apt install swh-plugins

• Select the whole presenter track
  • Effect -> Gate
  • Click OK
  • Check if the noise has been removed sufficiently from parts of the track where the presenter isn't talking. If not, Undo, and try Gate again adjusting the Threshold parameter.
  • Save

Equalisation

• Select all presenter tracks (not Samantha)
• Effect -> Equalisation
• Ensure Graphic EQ is selected, and move the first 6 sliders to the bottom Click OK
• Save

Remove peaks (optional)

If you're happy to spend some time to make it sound nicer, you can remove peaks before applying compression, which gives a more uniform overall result.

• Switch to the Envelope tool
• The track will have an inner band highlighted and a darker outer band behind the waveform. If the levels were set up correctly, most of the waveform should be inside the light area. If this is the case, identify any peaks that go into the dark area. If not, then just skip this step and go on to compression.
• For each peak,
  • Select the section of audio the peak is within (if you zoom in you'll see the shape of the syllable or noise)
  • Effect -> Amplify
  • On newer versions of audacity, you can set New peak amplitude to -6 or -7 and click OK
    • If not, set Amplification to -2 and click OK, then Effect -> Repeat amplify until it's reduced sufficiently. *Save

Editting (optional)

Listen through for the following:

• Someone talks over a more interesting/relevant point. Generate Silence over the talker overer. Yes, I made talker overer up.
• Someone (probably Martin) trips over their words and it can be easily editted out. Editted out the blunder.
• Generate silence over any obvious coughs or keyboard noise.
• Cut down any particularly lengthy silences.

Compression

Normalises volume, makes us sound sexier.

• For each presenter track (not Samantha) and don't be tempted to select them all at once as bad things will happen:
  • Select the whole track
  • Effect -> Compressor
    • Threshold -20dB
    • Noisefloor -30dB
    • Ratio 2.5:1
    • Attack Time 0.2 secs
    • Decay Time 1.0 secs
  • Wait...then do the next track.
  • Save

Everyone should now be at approximately the same volume, make any small adjustments needed to each track with the + - slider.

Panning

Leave Samantha dead center, move the other tracks left and right a bit with the L-R slider. My general principal is:

• The person who talks most one notches to the left
• The next most one to the right
• Next most two to the left
• Least two to the right. You might need to play around a bit. Afterwards, check each track is still audible in the ear it's not panned towards.

Give it a quick listen through and check there's nothing horrible. Check it's ok both with and without headphones - a lot of people listen in the car or around the house; they were the ones who complained when the voices too quiet. Save now.

Mix and render

• Select all tracks, Tracks -> Mix and render to new track.
• Select the new Mix track, Effect -> Compressor
  • Threshold -20dB
  • Noisefloor -30dB
  • Ratio 2:1
  • Attack Time 0.2 secs
  • Decay Time 1.0 secs
• If there's any other voice tracks from Samantha e.g. Interviews they might require amplyfying.
  • Highlight the interview (or whatever) from the beggining until the end of the talking, Effect -> Amplify, set New Peak Amplitutde to -4 and click OK
  • Have a test listen.
• Highlight the whole track (or just up to the end of the outro if you're leaving a "Stop Recording" at the end), Effect -> Truncate Silence.
• Set the Threshold to -40dB
• Set the Duration and Truncate to to 1s and click OK.
• Press Solo on the Mix track and press play to check you're happy with the result.
• Select the Mix track
  • File -> Export Selected Audio, save as a .flac file. Don't enter any tags, these get added by podpublish. Name the file S09E00.flac, where 09 is the season number and 00 is the episode number.
• Drop the .flac file back into the episode sub-directory in Dropbox.

Test listen

Get another member of the team to test listen to be sure the audio is good and everyone is in sync. Listening to the first 3 and last 3 minutes should be sufficient.