Local transcription for Windows 10 and 11
Transcribe interviews and meetings without uploading them.
Saypane turns recordings into verbatim transcripts on your own PC: timestamps, speaker labels and export to Word. For interviews you promised to keep confidential, patient conversations, client calls and anything else that should not go to a transcription service.
The free version transcribes the first 15 minutes of every file. Free download with free, unlimited dictation. Saypane Pro is a one-time in-app purchase of $24.99.
Made for recordings that must stay private
Researchers and PhD students
Consent forms and ethics approvals often limit where interview data may go. With Saypane it stays on the PC where you analyze it.
Journalists
Protect sources by keeping their voices off third-party servers. Find quotes by timestamp.
Lawyers
Client conversations and hearings transcribed on your own machine, with no processor to add to your data protection paperwork.
Doctors and therapists
For policies that do not allow patient audio to leave the device. Check your organization's rules; Saypane gives you the local part.
What you get
- Verbatim text. What was said, in order. No AI summary, no paraphrasing.
- Timestamps on every segment, to find the passage in the recording.
- Speaker labels (Pro): Speaker 1, Speaker 2 and so on, so questions and answers are easy to tell apart.
- Meeting recording (Pro): your microphone and the sound of your PC together, so both sides of a call are captured. No bot joins the meeting.
- Export to TXT for free, and to SRT, VTT, DOCX for Word, and JSON with Pro.
- Batch queue (Pro): drop in a folder of interviews and let it run.
How it works
-
Add a recording
Drag files onto Saypane, or start a meeting recording, and pick the language that is spoken.
-
Let it run on your PC
Parakeet handles 25 European languages quickly on an ordinary CPU. Whisper covers the rest. A GPU makes both faster.
-
Check and export
Check the transcript against the recording, then export to Word, subtitles or plain text.
Long files without loops
A known weakness of Whisper is that on long recordings it can get stuck and repeat the same sentence for minutes, or skip a whole passage. In our tests, decoding an entire file in one pass produced exactly that.
Saypane does not decode a file in one pass. It splits the audio at pauses with a voice activity detector, transcribes each part on its own, checks the result for repeated text and retries any part where it finds a loop.
Files it reads
MP3, M4A, WAV, AAC, WMA, MP4, MOV and other common audio and video formats that Windows can play. Video files are read for their sound track. Decoding uses Windows Media Foundation, which is part of Windows.
Local transcription apps on Windows
| Saypane | WizWhisp | Buzz | Vibe | |
|---|---|---|---|---|
| Runs on your PC | Yes | Yes | Yes | Yes |
| Speaker labels | In Pro | No, its listing says so | Not listed | Yes |
| Records PC sound for meetings | In Pro | No, files only | Not listed | Yes |
| Dictation into other apps | Yes, free | No | No | No |
| Price | Free 15 minutes per file, Pro one-time | Free small models, Pro $9.99 once | Free, open source | Free, open source |
Other apps as described on their websites, store listings and GitHub pages on 27 October 2026. Buzz and Vibe are free; if they do what you need, use them. Best offline transcription software for Windows. All comparisons
Good to know
- Speaker labels are automatic and can be wrong with crosstalk, similar voices or one distant microphone. Check them before you quote anyone.
- Speech models tend to leave out filler words such as "um" and false starts. If your method needs them, as conversation analysis does, add them while checking.
- Saypane does not summarize, translate or answer questions about your recordings.
- Saypane keeps processing local. Whether that is enough for your ethics board, GDPR assessment or HIPAA policy is for them to decide; this guide lists what to tell them.
Questions
Can I transcribe interviews locally with speaker identification on Windows?
Yes. Saypane Pro transcribes on your PC and labels speakers (Speaker 1, Speaker 2, ...). Nothing is uploaded. Among free open source apps, Vibe also offers speaker labels. Saypane and Vibe compared.
What does the free version transcribe?
The first 15 minutes of every file, as a complete, exportable transcript without a watermark. There is no limit on the number of files. Longer files, speaker labels, meeting recording and the other export formats need Saypane Pro. Free download with free, unlimited dictation. Saypane Pro is a one-time in-app purchase of $24.99.
How long does a one-hour interview take?
It depends on the model and your PC. On our test PC, Parakeet ran at about 13 times real time on 4 CPU threads, which is under 5 minutes for an hour of English, German, French or another of its 25 languages. Whisper large v3 turbo on a CPU alone ran at roughly real time or slower, so an hour can take an hour or more; a GPU through Vulkan roughly doubles that speed. Details.
Does it work without internet?
Yes, once the speech models are downloaded. You can disconnect the PC from the network and transcribe as usual.
Transcribe your first file
Free for the first 15 minutes of every file. No account, no upload.