Transcribe the source recording
Use the recommended model first. Noisy field recordings may benefit from a higher-accuracy option on capable hardware.
Interview transcription
Interview transcription you can verify against the recording
Turn research interviews, reporting calls, oral histories and customer conversations into timestamped text. Search for a topic, click back to the exact recording, correct important details and export a clean transcript.
- Search long interviews instead of scrubbing the whole recording
- Verify quotes and names from the original audio
- Keep sensitive source audio on the current device
Drop an audio file or click to browse
MP3, WAV, M4A, OGG, FLAC, WebM — practical length depends on device memory
Review the transcript before you export
A safer interview review process
Find claims and quotations
Search the text, click the timestamp and listen to the surrounding audio before relying on a quote.
Clean the deliverable
Correct names, organizations, numbers and terminology, then export text for notes, coding or publication.
interview transcription tool
Turn an interview recording into evidence you can revisit
Interview transcription is not finished when the words appear on screen. The valuable step is finding a claim, replaying its context, and correcting the details that can change what a source meant.
Find themes before you write
Search the transcript for a topic, product name or repeated phrase. Timestamps let you jump to the original answer instead of relying on a shortened line copied out of context.
For research and reporting, keep the recording beside the edited transcript. The transcript is a navigation layer and working draft, not an automatic publication-ready quote sheet.
Verify quotes, names and numbers
Replay the sentence before and after any quote you plan to use. Check names, organizations, dates and figures against the audio, especially when the speaker is far from the microphone or several people overlap.
EarScribe does not automatically identify speakers. Add speaker labels during your review when the distinction matters to the final document.
Export a working transcript for the next step
Use TXT for a reading draft, JSON when you need segment timing for research tooling, or SRT/VTT when the interview is also becoming a caption track.
Local processing reduces the need to upload an unpublished interview, but your device, browser and exported files still need the controls required by your organization or research protocol.
Review the transcript before you export
Before an interview quote leaves your notes
- 1
Locate
Search the exact phrase and save its timestamp before copying it.
- 2
Listen
Replay the surrounding answer so a short quote does not lose its context.
- 3
Correct
Verify names, organizations, numbers and specialist language manually.
- 4
Label
Add speaker names yourself when the transcript needs them; automatic labels are not provided.
Interview transcription needs human judgment
- The model may miss interruptions, accents, names and domain terms. Treat the output as a searchable draft that points you back to the source.
- Do not use local processing as a promise of absolute confidentiality. Follow consent, retention and device rules for the interview you recorded.
What interview transcripts still need from a human
Speaker identification is not automatic yet, so add names where the distinction matters.
Treat punctuation as a reading aid, not evidence of the speaker's exact phrasing.
Keep consent, retention and publication rules separate from the transcription tool itself.
Questions about this workflow
Can EarScribe identify different speakers?
Not yet. The transcript keeps timestamps, but speaker diarization and automatic speaker labels are not currently provided.
Should I publish an interview transcript without checking it?
No. Always verify quotations, names, numbers and statements that could change meaning or attribution.
Is interview audio uploaded?
The selected audio is processed in the browser and is not uploaded to an EarScribe server. See the Privacy page for model and analytics details.