The current API has limitations on how long an audio clip can be for transcription. It recommends that the audio file be cropped into shorter segments (https://docs.gladia.io/reference/limitations-and-best-practices#splitting-oversize-audio-files), which makes sense, though it’s a bit inconvenient when supplying URLs rather than uploaded audio files. It would be amazing if the API could support time_begin
and time_end
parameters to indicate that only a subset of the audio file should actually be transcribed!