DocsSessions and recording
Transfer and merge
How the files move to the host once the recording ends, and the details of the process that merges the voices into one file.
Transfer
When the recording stops, each guest sends their own file to the host.
- The file goes directly from computer to computer; it is not stored on the Mixapod server. On networks where a direct connection can’t be made, the encrypted traffic passes through a relay server, and it still isn’t stored.
- A ring that fills up with a percentage appears on the guest’s screen; when it’s done, it says Uploaded.
- The file is sent in 64 KB chunks. If the connection drops, the transfer resumes where it left off instead of starting over.
- The received file is verified with SHA-256: the file that reaches the host is bit-for-bit identical to the guest’s recording. If it doesn’t match, the guest sends the file again.
The host’s Mixapod requests missing files on its own: when collecting starts and when a guest reconnects. When everything has arrived, the merge starts on its own.
Merge
When everyone’s file has arrived, the host’s computer starts merging. The stages are shown on screen with a progress bar.
1. Alignment
The start frame of each file on the shared clock is known; the files are placed on the timeline with sub-millisecond precision. Several files from someone who joined the recording midway are also put in their right places.
2. Sound card speed correction
Even when two sound cards both say “48 kHz”, their clocks don’t run at exactly the same speed; over a one-hour recording the difference can add up to seconds. Mixapod measures each recording’s real speed and corrects differences up to ±1000 ppm, so the voices still line up at the end of the episode. The correction is applied to people whose recording consists of a single file.
3. Voice balancing
Each person’s voice is brought to −20 LUFS using ITU-R BS.1770 measurement; gain is raised by at most +18 dB and lowered by at most −24 dB. Someone who talks loudly and someone who talks quietly are heard at the same level.
4. Stereo mix
The voices are spread slightly across the stereo field in the order people joined the session, which makes it easier for listeners to tell who is speaking. The spread is kept narrow so it isn’t distracting on headphones.
| Number of people | Placement (−1 left, +1 right) |
|---|---|
| 2 | −0.25 · +0.25 |
| 3 | −0.30 · 0 · +0.30 |
| 4 | −0.36 · −0.12 · +0.12 · +0.36 |
| 5 | −0.36 · −0.18 · 0 · +0.18 · +0.36 |
Soundpad effects go into the mix as a separate Effects track, in the center and at the level they were played; they are not balanced.
5. Level and limiter
The mix is brought to podcast level, −16 LUFS; a limiter with 5 ms of lookahead keeps the peak level below −1 dBFS.
Output files
In the take folder (Documents\Mixapod\<date> <episode title>\Take 1\):
| File | Contents |
|---|---|
<episode title> - Take 1.mp3 | Stereo, 192 kbps: ready to publish |
<episode title> - Take 1.wav | Stereo, 48 kHz, 24-bit: for editing and archiving |
Tracks\<name>.wav | Separate tracks aligned to the timeline, mono, one per person |
Tracks\Effects.wav | Soundpad effects, as a separate track |
On the Recording ready screen you can listen to the merged recording, go to the files with Show in folder, move on to production with Add to production, or record a new take with Back to lobby. You can open files with Open and move them to the Recycle Bin with the trash can icon.
What happens to the raw recordings
With the default setting, once the merge is completed with everyone, the raw recordings, the .json files and the Tracks folder are moved to the Recycle Bin; only the MP3 and WAV stay in the folder.
- If you want to keep the separate tracks and raw recordings for editing, turn on Settings › Preferences › Keep separate parts.
- If someone’s file is missing, nothing is deleted; you can merge again once the missing file arrives.
- Moved files can be restored from the Recycle Bin.