How to transcribe › onnx-wakeword

How to transcribe onnx-wakeword files

onnx-wakeword makes the recording — Whipscribe makes it text. Three steps, and the first preview is free with no signup.

The three steps

1

Get the audio out. Export or locate the captured audio from onnx-wakeword and upload it.

2

Drop it below. Upload the file right here — no signup for the instant preview. Long, multi-hour files are fine, and video uploads work too (we transcribe the audio track).

3

Take the text with you. Accurate, speaker-labeled text with timestamps — export TXT, SRT, VTT or DOCX, in 100+ languages, processed on Whipscribe's own private cloud.

About onnx-wakeword

“Offline keyword spotting & wake word engine. ONNX format, model <130KB, inference <10ms. Runs on Android, ESP32, Linux, Web. No cloud, no data usage.”

GitHub stars
59
Built in
C
Category
Voice Assistants

onnx-wakeword facts & alternatives →  ·  GitHub ↗  ·  Website ↗

Frequently asked

Can Whipscribe transcribe voice recordings from onnx-wakeword?

Yes — any audio or video file onnx-wakeword produces can be uploaded directly. Export or locate the captured audio from onnx-wakeword and upload it.

Do I need to convert the file first?

No. Common audio and video formats upload as-is; video's audio track is transcribed automatically.

How accurate is it?

Clear speech in major languages comes back near-publishable; noisy or heavily accented audio deserves a review pass. The instant preview shows real output before you commit.

What does it cost?

About $2 per audio hour as pay-as-you-go credits that never expire — the first preview is free with no signup.