Skip to main content

Recording & Segments

How AutoPod captures your station's audio and stores it as small, resilient segments.

How recording works​

AutoPod runs a dedicated recorder for each stream. The recorder:

  1. Connects to your audio stream (Icecast, Shoutcast, or any HTTP audio source)
  2. Captures the audio continuously using FFmpeg
  3. Splits the audio into small segments — approximately 40 seconds each
  4. Uploads each segment to cloud storage as soon as it's captured
  5. Logs the segment in the database with its precise timestamp and duration

The recorder runs 24 hours a day, 7 days a week. It doesn't start and stop with your schedule — it records everything.

Why segments?​

Splitting audio into small segments rather than long continuous files provides several benefits:

Resilience — if your stream drops or glitches for a few seconds, only one segment is affected. The rest of the recording is intact.

Flexibility — episodes can start and end at any segment boundary. If your schedule changes, AutoPod can slice the audio differently without re-recording anything.

Efficiency — segments are uploaded immediately, so there's no risk of losing hours of audio if something goes wrong. And the episode generator can download just the segments it needs, not the entire day's recording.

The 60-day archive — because segments are stored independently, they can be retained for 60 days and used to create episodes retroactively.

What happens if the stream drops?​

If your stream goes offline, the recorder detects the interruption and waits for it to come back. When the stream returns, recording resumes automatically. The gap is recorded in the segment index, so the episode generator knows about it.

The resulting episode will have a slightly shorter duration than scheduled — the gap is simply absent from the audio, not filled with silence.

info

Stream drops are relatively rare but do happen (network issues, encoder restarts, etc.). AutoPod handles them gracefully — you don't need to do anything.

Segment storage​

Segments are stored in S3-compatible cloud storage. Each segment is a small audio file (typically 200-500 KB) with a precise timestamp.

The segment index in the database records:

  • Stream ID — which stream this segment belongs to
  • Timestamp — the exact time the segment was captured (Unix epoch)
  • Duration — how long the segment is (in seconds)
  • Storage path — where the file lives in cloud storage

This index is what allows the episode generator to find exactly the right segments for any time window.

Retention​

Raw audio segments are retained for 60 days from the time they were recorded. After 60 days, the segment files are automatically deleted from cloud storage. The database records remain for a short time afterwards as an audit trail.

This means you have a 60-day window to create episodes from any part of your broadcast history.

warning

The 60-day retention applies to raw segments. Published episodes have their own retention period based on your AutoPod plan (30 days, 90 days, or unlimited).