Formats and quality

This page is about what the files are made of. It covers the audio format, the video resolutions and bitrates, screen sharing, the difference between the live picture and the recording, and how much space an hour of recording takes.

Audio¶

Every audio track is an uncompressed WAV at 48 kHz, 16 bit, mono. There are no quality tiers for audio, and every plan records it exactly like this.

It's uncompressed because any compression would be a decision someone makes for you before the edit. A compressed format can sound fine on its own, but it doesn't hold up under processing, and a podcast track gets processed. It gets cut, levelled, denoised, and compressed once more for the feed at the very end. So the WAV is your master, and the small file comes last.

It's mono because a voice is one source. A stereo file from a single microphone would just hold the same audio twice at twice the size. That's why podsink asks your microphone for one channel at 48 kHz and reads the raw samples straight from it, with no encoder in between.

Video¶

Video comes with the Free, Max and Business plans. The host picks a resolution for the session, from 720p to 2160p, and each participant's camera is recorded at that resolution, or at a lower one if they've picked it themselves. The picture goes into an MP4 as H.264, at a bitrate that depends on the resolution:

These are ceilings tuned for a talking head, so a real file usually lands a little under them. The bitrate follows the size that's actually being recorded rather than the session's ceiling, so a 1080p camera in a 2160p session records at 5 Mbit/s.

The MP4 also carries an AAC copy of that participant's audio. That way the video plays on its own and lines up neatly with the WAV in your editor. For the edit you still use the WAV, though. The AAC is just there for reference.

Screen share¶

A shared screen, window or tab is recorded as its own MP4 at up to 5 Mbit/s, at whatever size the source has right now. That size changes, though, as soon as someone resizes a window or switches displays. So podsink uses a variant of H.264 for this track that can change resolution mid-stream without breaking the file. The quality preset in the session, "Crisp" for text or "Motion" for video, tells the browser what to focus on.

Live is not the recording¶

What the others see and hear of you during the session goes through a media server, so you can all talk to each other. That stream is mainly meant to run smoothly. With up to three people in the room, the live picture comes in 1080p, with four to six in 720p, and with seven or eight in 540p. None of that touches the recording, because it's made on your device the whole time, at the resolution set for recording. It's the same with audio. The live audio is compressed for the call, the WAV isn't.

How big an hour gets¶

Here's how much space an hour of recording takes per participant, worked out from the bitrates above, in decimal gigabytes:

With more people and longer recordings, that adds up fast. A two-hour recording of four people at 1080p comes to around 21 GB, and each participant uploads their share of that over their own line while they talk.

Why it takes Chromium¶

Reading raw samples from the microphone and writing H.264 at these sizes into an MP4 needs browser features that so far only Chromium-based browsers have. That's why podsink supports Chrome and Edge. What's behind it, and what works on phones, is on Devices and browsers.