Use cases · Research interviews
Interview transcription API for user research and qualitative analysis
Speech Revolutions transcribes interviews and focus groups with the interviewer and each participant separated and every word timestamped, so research tools can tag quotes, link them to the moment in the recording and attribute them correctly. On interview audio it measured less than half the diarization error rate of AssemblyAI and Deepgram.
from speechrevolutions import SpeechRevolutions
client = SpeechRevolutions()
result = client.transcribe("participant-07.mp4")
for u in result.utterances:
print(f"[{u.start:7.1f}s] {u.speaker}: {u.text}")Where it goes wrong
Qualitative research runs on quotes. A participant's insight credited to the interviewer, or two participants merged into one, breaks synthesis: the theme is supported by the wrong person, or by one person instead of three.
Focus groups are the hard case: four to eight people, overlapping answers, and short turns like "same for me" that are easy to drop or misassign.
What the API gives you
- Speaker turns
- Utterances give one block per turn, ready to tag and quote.
- Quote timestamps
- Word-level start and end times link every highlight back to the exact moment in the recording.
- Many languages
- Language is detected automatically across 99+ languages, including participants who switch mid-answer.
- Exports
- DOCX and PDF transcripts for researchers who work outside your tool.
- Privacy
- Recordings and transcripts are deleted within 30 minutes of completion and never used for training.
Measured
On VoxConverse interviews and talk shows, Speech Revolutions measured 3.6% diarization error rate, against 9.8% for AssemblyAI and 10.6% for Deepgram.
Full results and method: speaker diarization measured across five APIs and all benchmarks.
Pricing
$0.003 per audio minute, $0.18 an hour, billed by the second. Speaker labels, word timestamps and every output format are included. New accounts get $10 of credit with no card required. Pricing details.