Introducing Transcribe 2.0: Safe for sensitive data
Accurate results, 2 to 5 times faster
Use your @brown.edu Google account
Interviewer 00:04
Thanks for making time today.
Student 00:07
Happy to. Where should we start?
Interviewer 00:11
Tell me about your first semester.
Speaker 1 00:03
Welcome back, everyone. Let's go around the room.
Speaker 2 00:08
Hi, I'm [first_name] [last_name]. I moved here from [city] last fall.
Speaker 1 00:15
Thanks. Feel free to call me at [phone_number].
Secure · Accurate · Fast
Built by Brown, for the Brown community
Free for all Brown faculty, students and staff. Up to about 20 hours of audio or video a day, resetting at midnight ET.
- users trust Transcribe
- 600+
- hours transcribed
- 5,000+
- to process 1h of media*
- ~30s
- languages supported
- 16
*Using Cohere Transcribe model without speaker detection
Features
Professional-grade transcription, a few clicks away
Transcribe does more than run a model. Extra steps before and after transcription, drawn from the latest speech-to-text research, give you fewer hallucinations, precise timestamps* and reliable speaker labels.
*Timestamps may be less precise with the Gemini model
Smart recommendations
Choose from 4 state-of-the-art models, or let Transcribe recommend one for you.
Current lineup
- Google GeminiFlash
- OpenAI WhisperPro
- Alibaba Qwen3-ASRPro
- Cohere TranscribePro
Drop in any file
Audio or video, in any popular format. Upload it and get a transcript back in seconds or minutes.
Some supported formats
- Audio:.wav, .mp3, .m4a, .ogg, .flac …
- Video:.mp4, .mov, .webm, .m4v …
Up to 5 hours and 1 GB per file
Review, edit and export
Click any line to hear that moment, fix words and rename speakers, then download as Word or plain text.
Word-level timestamps in JSON also available with Pro models
Captions and subtitles
Improve the accessibility of lectures and videos. Preview subtitles, then download them as SRT or VTT to check and publish.
Available with Pro models
Translation
Upload a recording in one language and get the transcript in another, across 15 languages.
Available with the Gemini model
Shaped by the community
Most new features start as a request from someone at Brown.
In the works
- De-identificationExperimental
- Panopto integrationComing soon
Responsible AI
We care about privacy and sustainability
Transcribe is designed for privacy, security and sustainability. Its servers shut down when no one is using it.
Approved for Level 3, Brown's highest data risk classification
Transcribe can process sensitive information, including protected health information (PHI) and FERPA-protected student records. See what qualifies as Level 3 in Brown's data risk classifications.
- Your data is never used to train AI
- Your audio and video files are deleted within 7 days of upload
- You can delete your transcripts at any time
We track and reduce Transcribe's carbon footprint
Unlike many AI tools, Transcribe shows you the estimated carbon footprint of each job. We choose models for efficiency as well as accuracy, which cuts processing time and emissions.
Frequently Asked Questions
Before you upload
What researchers and the rest of the Brown community ask most often.
All questionsCan I use my research data with Transcribe?
Transcribe is approved for Level 3 data and below, under Brown's data risk classifications. Check your data's level, and if it's human-subjects research, get approval from the IRB/HRPP office before you upload.
How long are my files kept?
Audio and video files are stored for up to 7 days, then deleted permanently. Once a transcription is done, you can delete the files and transcripts yourself at any time. See the Transcribe data privacy policy.
Can anyone else access my data?
No. Your files and transcripts are private to your Brown account and are never shared with third parties.
Are there any limits to using Transcribe?
Each person can transcribe up to about 20 hours of audio or video a day, resetting at midnight ET. A single file can be up to 5 hours long and 1 GB in size.
