turian / whisply

Transcribe, translate, annotate and subtitle audio and video files with OpenAI's Whisper ... fast!

Cold

Public
27 runs
T4
GitHub
License

Iterate in playground

Run with an API

Playground API Examples README Versions

Input

Run this model in Node.js with one line of code:

npx create-replicate --model=turian/whisply

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run turian/whisply using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "turian/whisply:770f8aa6121dbf8dca689cde6343a54dccf1f44d105ffd1711cff4ef4ac007c7",
  {
    input: {
      model: "distil-large-v3",
      verbose: false,
      annotate: false,
      subtitle: false,
      translate: false,
      sub_length: 5
    }
  }
);

// To access the file URL:
console.log(output.url()); //=> "http://example.com"

// To write the file to disk:
fs.writeFile("my-image.png", output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run turian/whisply using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "turian/whisply:770f8aa6121dbf8dca689cde6343a54dccf1f44d105ffd1711cff4ef4ac007c7",
    input={
        "model": "distil-large-v3",
        "verbose": False,
        "annotate": False,
        "subtitle": False,
        "translate": False,
        "sub_length": 5
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run turian/whisply using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "turian/whisply:770f8aa6121dbf8dca689cde6343a54dccf1f44d105ffd1711cff4ef4ac007c7",
    "input": {
      "model": "distil-large-v3",
      "verbose": false,
      "annotate": false,
      "subtitle": false,
      "translate": false,
      "sub_length": 5
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

Output

No output yet! Press "Submit" to start a prediction.

Run time and cost

This model runs on Nvidia T4 GPU hardware. We don't yet have enough runs of this model to provide performance information.

Readme

whisply-replicate

A Replicate.com service for audio transcription, translation, and speaker diarization using Whisper v3 models.

Features

Multiple Whisper Models: Support for various Whisper models including:
large-v3 (default)
distil-large-v3
large-v3-turbo
And many more standard Whisper models
Advanced Audio Processing:
Automatic audio format conversion and normalization
Support for various input audio/video formats
GPU-accelerated processing
Rich Output Options:
Basic transcription (txt, json)
Subtitle generation (srt, vtt)
Speaker diarization with word-level timestamps
Translation to English

Usage

The service accepts the following parameters:

Required:

audio_file: Input audio/video file to process

Optional:

language: Language code (e.g., ‘en’, ‘fr’, ‘de’) [default: auto-detect]
model: Whisper model to use [default: ‘large-v3’]
subtitle: Generate subtitles (.srt, .vtt) [default: false]
sub_length: Words per subtitle segment [default: 5]
translate: Translate to English [default: false]
annotate: Enable speaker diarization [default: false]
num_speakers: Number of speakers to detect [default: auto-detect]
hf_token: HuggingFace token for speaker annotation
verbose: Print progress during transcription [default: false]
post_correction: YAML file for text corrections

Example Usage with Cog:

# Basic transcription
cog predict -i audio_file=@path/to/audio.mp3

# Full features with speaker diarization
cog predict -i audio_file=@path/to/audio.mp3 \
           -i language=en \
           -i model=large-v3 \
           -i subtitle=true \
           -i translate=true \
           -i annotate=true \
           -i hf_token=your_token_here \
           -i num_speakers=2

Output

The service returns a zip file containing: - Transcription in requested formats (txt, json) - Subtitle files if requested (srt, vtt) - Speaker annotations if enabled (rttm format) - Translated text if translation was enabled

Technical Details

Uses FFmpeg for audio preprocessing
Automatic GPU detection and utilization
Persistent model caching for faster startup
Error handling and validation for all inputs
Support for various audio formats through python-magic detection

Model Caching

Whisper v3 models are pre-downloaded during container build
Speaker diarization models (when using annotate=true):
Require a valid HuggingFace token
Are cached after first use
Use persistent storage for subsequent runs