MP4 video has no sound? Find where the audio went

If an MP4 video has no sound, first find out whether the file has an audio track at all. With no track, the file was made without sound, and you fix the step that made it; with a track, either the track itself is silent or the player can't play it. For a file on Sume, a free video inspect probe answers the first question with has_audio and names the track's codec.

The facts come from the Video inspect, Video trim, Timeline 1.0, Timeline compose, Video filter, and Video generation docs and the Sume API reference, read on 2026-09-27. Anything described as current behavior is read from Sume's code.

How do I check whether an MP4 has an audio track?

Run a probe-only inspect: POST /v1/video-inspect with frames: false. It is unbilled and never re-encodes the source, and the default sync mode answers within 30 seconds or returns a job to poll. The clip must be your workspace's media.sume.com file, such as an earlier Sume job's output; which URLs each endpoint accepts explains the rule.

• probe.has_audio is false: the file has no audio track. Find the step that made it, below.
• has_audio is true: audio_codec, audio_channels, and audio_sample_rate describe the track. No probe field measures loudness, and in current code has_audio is true whenever the file has an audio stream, even a silent one.
• If the same file plays with sound in another player, the file is fine, and the problem is the player or the device.

curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: probe-silent-clip-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/clip.mp4",
    "frames": false
  }'

Which Sume steps leave a video without sound?

Each of these steps writes a new file, so the fix is to redo the step with the sound kept, not to repair the output.

Can Sume bring the sound back?

Only if the sound still exists somewhere. Trim, filter, and detach never change their source, so when an earlier step dropped the sound, run that step again from the original. For a Timeline render, detach the sound with POST /v1/audio-detach, whose default wav is the file a spine wants, and render again.

A track that was never recorded or generated can't be recovered: detach fails with detach_source_has_no_audio, and an inspect transcript fails with inspect_source_has_no_audio. Add a new track instead by rendering the clip over narration or music, as in add a voiceover to a video. Which video models generate audio lists the models that make sound.

What if the MP4 has an audio track that won't play?

Check audio_codec. A trim of the whole clip at the default exact precision re-encodes the kept audio as AAC (at 192 kbps in current code), while keyframe precision is a stream copy that keeps the audio as it was. Trim takes sources up to 1,800 seconds and writes at most 900 seconds per job. It is billed per job, plus a 5.5% agent fee by default; the docs say to confirm the rate in GET /v1/catalog.