Distil-Whisper Large v3

Distilled Whisper — several times faster than large-v3 for English with close accuracy.

Overview

Distil-Whisper keeps Whisper's encoder and shrinks the decoder, giving a large speed-up for English transcription.

Key features

• ✓
  Much faster than large-v3
• ✓
  English-focused
• ✓
  Same API

Use cases

• →
  High-volume English transcription
• →
  Real-time captioning experiments

On this template

Runs on the Whisper API template, exposing its Transcription API at a private HTTPS address. You also get root SSH and JupyterLab.

WHISPER_MODEL

distil-whisper/distil-large-v3

Quick start

1. Click Deploy and pick an available GPU — only offers with enough VRAM are shown.
2. Wait for the status to turn running. The first start downloads the model, which can take a few minutes.
3. Use the connection details on your rental page:

curl https://r12-abcd1234-app.gpu.clearlayer.ca/v1/audio/transcriptions \
  -H "Authorization: Bearer <app-token>" \
  -F file=@meeting.mp3 -F model=whisper-1

from openai import OpenAI

client = OpenAI(base_url="https://r12-abcd1234-app.gpu.clearlayer.ca/v1", api_key="<app-token>")
with open("meeting.mp3", "rb") as f:
    print(client.audio.transcriptions.create(model="whisper-1", file=f).text)