The quality of your transcript is decided before the first word is typed – namely, during the recording. Good sound makes transcription faster, more accurate and cheaper; poor sound makes it more expensive and lets errors creep in. The good news: for a clean recording you do not need a recording studio, just a few simple steps. Here are eight practical tips and the answer to the question of the right file format.
Why does sound quality determine price and accuracy?
Every bit of noise, every echo and every quiet passage forces the typist – or the AI – to rewind and guess. With clean sound, the transcription runs smoothly; with poor sound, the time required rises considerably, and with it the price. At the same time, accuracy drops, because unintelligible passages have to be marked as gaps. In short: ten minutes of care during the recording saves hours and real money later.
The 8 most important tips for a better recording
- Choose a quiet room: closed windows, no street noise, no open-plan office. Textiles such as carpets and curtains absorb echo.
- Use an external microphone: a simple clip-on or USB microphone clearly beats any built-in laptop or phone microphone.
- Keep the right distance: the microphone belongs close to the speakers – about a hand's width away. Too far away means thin, echoey sound.
- Make a test recording: record 30 seconds and listen back before you start. That way, background noise stands out while it can still be fixed.
- Only one person speaks at a time: ask participants not to interrupt each other. Overlaps are the most common reason for gaps in the transcript.
- Switch off sources of noise: ventilation, air conditioning, radio and notification sounds off. Phones on silent, not on vibrate.
- Record in a suitable file format: uncompressed WAV or a high-quality MP3 – more on this below.
- Check battery and storage: charge and free up space before the appointment. An interrupted recording cannot be made up for.
Which file format is best for transcription?
The best choice is an uncompressed WAV, because it stores the recording losslessly and preserves every detail. If the file is too large, an MP3 with a high bit rate (192 kbps or more) is a good compromise between quality and file size. Above all, it is important not to compress twice – in other words, not to convert an MP3 that is already heavily compressed once again, because sound information is irretrievably lost in the process.
- WAV: lossless, best quality – ideal if storage and upload are not an issue.
- MP3 (192 kbps or more): a good compromise, small enough for a convenient upload.
- M4A/AAC: also usable, provided it is recorded at a high bit rate.
- Avoid: heavily compressed voice notes with a low bit rate and double conversions.
What can you do if the recording is still poor?
Not every recording situation can be planned perfectly – sometimes the conversation took place long ago. Even then, not everything is lost: an experienced typist can often still get a surprising amount out of difficult material, because they interpret unclear passages in context and listen to them several times in a targeted way. Our comparison automatic vs. manual transcription shows what a human achieves here and where automatic tools fail with poor sound.