Audio and video formats for transcription

Ce contenu est en anglais.

Transcription depends on the sound more than the picture, so the format you choose matters less than the quality of the recording inside it. Still, some formats are easier to upload, some sound better at a given file size and some are better kept as originals. This guide explains the common audio and video formats you are likely to meet, how mono and stereo recordings differ, how bit rates affect file size, and how to keep your originals intact while preparing a file for the Video Transcriber on Valtrinity.

Par Valtrinity. Mis à jour le 18 octobre 2026.

The formats you will meet

MP3 is the most common compressed audio format and is accepted almost everywhere. M4A is a compressed audio format often produced by phone voice recorders. WAV is uncompressed and gives a faithful copy of the sound, at the cost of very large files. FLAC is a lossless compressed format, so it keeps the full sound while taking less space than WAV.

For video, MP4 is the most widely accepted container, and MOV is common from cameras and editing software. WebM is an open web video format that is common on the internet. A video file contains both picture and sound, and the transcriber only needs the sound to produce a transcript.

• MP3: compressed audio, accepted almost everywhere.
• M4A: compressed audio from many phone recorders.
• WAV: uncompressed, faithful and large.
• FLAC: lossless and smaller than WAV.
• MP4, MOV, WebM: video containers that include sound.

Mono or stereo for speech

A mono recording has one channel and a stereo recording has two. For speech, mono is usually enough. A single microphone produces mono sound, and most transcription depends on the voice rather than on where it sits in the stereo field. Stereo files are larger, and if the two channels contain the same sound, the extra channel adds size without adding detail.

Stereo is worth keeping when the recording has two separate sources, such as two microphones in different places or a presenter and a separate source on the other side. Even then, check the file, because some recorders put the same sound on both channels.

• Mono is usually enough for a single speaker or a room.
• Stereo adds size without detail if both channels match.
• Keep stereo when the sources really are separate.

Bit rates and file size

The bit rate of a compressed file sets how much data is used each second. A higher bit rate keeps more of the original sound and produces a larger file. For speech, a moderate bit rate is usually enough, and extra quality is rarely needed for transcription. Music and recordings with many different sounds benefit more from higher rates.

File size depends on both length and quality. A long recording at a high bit rate can be larger than a short video. Check the size against the upload limit, which is up to 500 MB for the Video Transcriber, before you start.

• Higher bit rates keep more detail and produce bigger files.
• Speech needs less than music.
• Check file size against the 500 MB limit.

Worked example: a long recording

Suppose you have a three-hour recording of a conference session saved as an uncompressed WAV file. It is far larger than the 500 MB limit, so it cannot be uploaded as it is. Converting it to a compressed MP3 at a moderate bit rate keeps the voices clear and makes the file much smaller. Upload the MP3 and keep the WAV as your original.

Note that a single file pass covers up to 90 minutes for £0.99, and the Unlimited plan includes 600 minutes a month with files up to 180 minutes. A three-hour recording should be split into parts that fit those limits, so check the length before you convert or split it.

• Convert large WAV files to MP3 for speech.
• Keep the WAV original.
• Check length against the 90 minute and 180 minute limits.

Keep your originals

Every conversion loses something, even if the loss is hard to hear. Keep the original recording in its best format and work on copies. If you convert a compressed file into another compressed file, the quality can drop each time, so convert from the original rather than from a previous copy.

Name the original and the copy clearly, such as interview-original.wav and interview-upload.mp3, so you never upload the wrong one or delete the only good copy.

• Keep the best-quality original.
• Convert from the original, not from an earlier copy.
• Name originals and copies clearly.

Choose a format and upload

For most transcription jobs, a clear MP3 or MP4 at a moderate quality is the simplest choice. Upload it to the Video Transcriber on Valtrinity, turn on speaker labels if you need them, and check the first minute of the transcript for quality. If the first minute is unclear, the problem is usually the recording, so fix that before uploading the rest.

• Use MP3 or MP4 for most jobs.
• Check the first minute before continuing.
• Fix the recording rather than correcting a poor transcript.

Outils cités dans ce guide

Questions fréquentes