Choose a recording
Add a supported audio or video file, or record directly from your microphone.
FREE SPEAKER DIARIZATION · PRIVATE
Transcribe conversations with speaker labels.
Anonymous, local, and free.
Separate different voices as Speaker 1, Speaker 2, and more while your recording remains on your device.
✓ Files stay on your device✓ 99 supported languages✓ TXT & subtitle export
MULTILINGUAL SPEECH TO TEXT
Transcribe 99 supported languages
Choose from the complete Whisper language list or let the tool detect the spoken language automatically. Support includes English, Spanish, Chinese, Arabic, Hindi, Japanese, Russian, Vietnamese, and many more.
WHISPER AI TRANSCRIPTION
Free Whisper transcription in your browser
This tool uses the open-source Whisper speech recognition model to transcribe audio and video locally. LocalScribe is not affiliated with OpenAI. Learn about Whisper transcription →
01
Upload audio or video
Your file stays private and is processed on your device.
or
02
Choose your transcription settings
Fast transcription is selected by default.
◇Ready when you are
HOW IT WORKSGet an editable transcript in three simple steps.
Get an editable transcript in three simple steps.
No account required.
Start transcription
Select the spoken language or use automatic detection, then follow the progress on screen.
Edit and download
Correct the text and save it as TXT, SRT, or WebVTT subtitles.
WHO SPOKE WHEN
Make multi-speaker recordings easier to follow
Optional speaker diarization analyzes voice characteristics and aligns anonymous labels with the timestamped transcript. It does not recognize names or identities.
Meeting speaker labels
Distinguish participants in recorded calls and discussions when reviewing notes and decisions.
Interviewer and guest turns
Separate questions and answers into an easier-to-read interview transcript.
Multi-host podcast transcripts
Label changing voices in episodes, roundtables, and recorded conversations.
Labeled TXT, SRT, and VTT
Download speaker-prefixed text and timestamped subtitle cues after reviewing the result.
Popular uses and formatsMeeting transcriptionInterview transcriptionPodcast speakersSpeaker-labeled SRTSpeaker-labeled VTT
Speaker analysis without uploading your recording
Your selected recording and completed transcript stay on your device. The transcription tool does not upload them to our servers.
Speaker diarization questions
What is speaker diarization?
Speaker diarization estimates who spoke when. This tool uses anonymous labels such as Speaker 1 and Speaker 2; it does not identify people by name.
Is speaker diarization processed locally?
Yes. When enabled, both transcription and speaker analysis run on your device. Additional speaker models are downloaded and cached by the browser the first time.
How accurate are the speaker labels?
Quality depends on the recording. Overlapping speech, background noise, short replies, and similar voices can cause mistakes, so review labels before publishing.
Why does speaker labeling take longer?
The optional feature performs separate voice segmentation and comparison after transcription. Its extra models and processing are only used when you enable speaker labels.