When you pull a quote from a customer interview, you need to know who said it and where to find it in the audio. We ran the same two-person recording through ElevenLabs Scribe v2, Otter and faster-whisper to see what you can use straight from the export and what you need to check.
Two synthesized English voices discuss a fictional startup’s research process. The recording has ten turns and a brief overlap around 0:46. Each tool received the same WAV, with no name hints.
At 0:30 the interviewer asks who spotted the difference. Only Scribe kept the name intact in both mentions. All three got Nia Okafor and Cleo Varela right, and all three misspelled Qivora.
Reference
Was it Amira Nwosu who spotted the difference?
Scribe
Was it Amira Nwosu who spotted the difference?
Otter
Was it Amira and Nwosu who spotted the difference?
faster-whisper
Was it Amir and the wasu who spotted the difference?
All three kept 14 distinct from 40 and preserved the $2,490 quote, $2,940 invoice and $49.90 export fee. They also captured 2.6% and 3:15 p.m. UTC. Otter wrote “$49.90” where Scribe wrote “49 dollars and 90 cents.”
At 0:53 the participant asks for “A clean text file.” Otter attached “A” to the interviewer’s question. Check both sides of a speaker change before quoting either person.
Compare the raw Otter and Scribe TXT exports
Otter TXT
Speaker 1 0:50 Got it. What would make the export worth paying for? A
Speaker 2 0:53 clean text file with timestamps and a way to see who said each thing. We need to find the original quote again before it goes into a research report.
Scribe TXT
00:00:50,380 --> 00:00:53,260 [Speaker 0] Got it. What would make the export worth paying for?
00:00:53,680 --> 00:01:02,800 [Speaker 1] A clean text file with timestamps and a way to see who said each thing. We need to find the original quote again before it goes into a research report.
Try it on one of your interviews
This short sample shows where to look for mistakes. Use a recording you have permission to process to see how a tool handles your interviewees’ accents, microphones and interruptions.
Pick a minute with a tricky name, a figure and a change of speaker.
Open the export where your team will use it. Check that names, speaker labels and timestamps come through.
Replay one quote from its timestamp. Check the wording and who said it, then keep the timestamp with the quote in your notes.
Disclosure: Some product links are affiliate links. We may receive a commission from a qualifying purchase. Personal use, documented tests and source-based assessments are identified where stated; inclusion alone does not mean a product was personally tested.