Transcribing WAV recordings
Why WAV files are large, and why that is fine
WAV stores audio uncompressed: an hour of CD-quality stereo is about 600 MB, ten times an MP3 of the same recording. On a computer you can upload WAV files of up to 1 GB, and they are shrunk to speech-quality audio before upload, so a big file does not mean a long upload.
The extra detail of WAV does not hurt accuracy, and it does not raise it much either: speech models listen at speech quality. What matters is a clear voice, close to the microphone.
Common sources of WAV audio
- Field and handheld recorders (Zoom H-series, Tascam, Sony)
- Digital dictation devices used in legal and medical offices
- Audio exported from Audacity, GarageBand, Adobe Audition or a DAW
- Call recordings and voicemail exports from phone systems
- Wireless lavalier kits that record a backup track
Tips for the cleanest transcript
If the recorder made a separate track for each microphone, upload the track of the main speaker, or a mix in which every voice is clear. Trim long silences at the start if you like, although it is not needed. Choose the spoken language when you upload, for the best result on names and terms.
TranscriptGo does not label speakers. Lines are timestamped, so you can always check who said what against the audio.