skryba.ai

AAC to text

Audio encoded the way it is inside an M4A, only served without that wrapper — which is where the odd player behaviour comes from.

Ready to transcribe

Check the language and we'll begin.

Record now

MP3, M4A, WAV, MP4, MOV and more · up to 5 GB · or drop a file anywhere

No account needed. Recordings up to 5 minutes are free.

Recording language

Leave “Detect” on unless the conversation mixes Polish and English.

Checking the length…

  1. Upload
  2. Transcription
  3. You read
  1. Uploading the file
  2. Transcription
  3. The beginning, ready to read

Encrypted upload · automatic AI transcription

In short

You convert an AAC file to text by uploading it without conversion. A bare .aac is usually an ADTS stream rather than a container, so nothing in its header states the duration — which is why a player shows the wrong time or stumbles when seeking. The file is fine, and we read it start to finish regardless of what the progress bar says.

How to do it

  1. Upload the AAC without converting it, even if your player shows an odd duration for it.
  2. Leave language detection on automatic, or pick Polish or English.
  3. Wait for processing — an hour of audio usually takes a few minutes.
  4. Read the preview; unlock the full transcript with a one-off payment or subscription hours.

AAC and M4A — codec versus container

AAC is a way of encoding sound. M4A is the box that same encoded sound usually sits in — along with a title, a duration and the rest of the metadata. A file ending in .aac is most often that same audio with no box: a bare ADTS stream, frames glued one after another.

  • A player does not know the length up front, so it estimates — and with a variable bitrate it can be minutes out.
  • Seeking can be imprecise, because there is no index to jump into.
  • Concatenating two .aac files by plain copying simply works, which is not true of containers.

None of that gets in the way of transcription. The timestamps in the finished transcript are computed from the audio itself, so they are correct even when the player on your computer insists otherwise.

Where .aac files come from

  • Recorders and cameras that write the audio track separately from the picture.
  • Captured broadcasts and radio streams, where ADTS is the transmission format.
  • Exports from some editing software, when “raw AAC” is picked instead of M4A.
  • Voice notes from some messaging apps — Messenger and Signal hand over AAC or M4A.

In every one of those cases, upload the file without converting it. Transcoding AAC to MP3 is re-encoding lossy into lossy — the file grows or loses quality, and the recognition result gains not a single correct word.

Upload an AAC file

Questions

My player shows the wrong duration for the AAC

That is normal for a raw ADTS stream — it has no header stating the length, so the player estimates it. The file is fine, and the transcript's timestamps are computed from the audio itself, so they will be correct.

How does AAC differ from M4A?

The codec is the same; the wrapper differs: M4A is a container with metadata, while .aac is usually a bare stream without any. We take both identically and read the same audio from each.

Will converting AAC to MP3 improve the transcript?

No — it will make things worse, or at best change nothing. It is re-encoding one lossy format into another, which adds artefacts rather than information. Upload the original.

Got a recording? In a few minutes you'll read it as text.