vaibhavs10 / incredibly-fast-whisper

whisper-large-v3, incredibly fast, powered by Hugging Face Transformers! 🤗

Warm

Public
5.7M runs
L40S
GitHub
License

Run with an API

Playground API Examples README Versions

Input

Video Player is loading.

Current Time 00:00:000

Duration 00:00:000

Loaded: 0%

Stream Type LIVE

Remaining Time 00:00:000

audio

*file

Audio file

task

string

Task to perform: transcribe or translate to another language.

Default: "transcribe"

language

string

Language spoken in the audio, specify 'None' to perform language detection.

Default: "None"

batch_size

integer

Number of parallel batches you want to compute. Reduce if you face OOMs.

Default: 24

timestamp

string

Whisper supports both chunked as well as word level timestamps.

Default: "chunk"

diarise_audio

boolean

Use Pyannote.audio to diarise the audio clips. You will need to provide hf_token below too.

Default: false

hf_token

string

Shift + Return to add a new line

Provide a hf.co/settings/token for Pyannote.audio to diarise the audio clips. You need to agree to the terms in 'https://huggingface.co/pyannote/speaker-diarization-3.1' and 'https://huggingface.co/pyannote/segmentation-3.0' first.

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

npx create-replicate --model=vaibhavs10/incredibly-fast-whisper

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 vaibhavs10/incredibly-fast-whisper using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "vaibhavs10/incredibly-fast-whisper:3ab86df6c8f54c11309d4d1f930ac292bad43ace52d10c80d87eb258b3c9f79c",
  {
    input: {
      task: "transcribe",
      audio: "https://replicate.delivery/pbxt/Js2Fgx9MSOCzdTnzHQLJXj7abLp3JLIG3iqdsYXV24tHIdk8/OSR_uk_000_0050_8k.wav",
      language: "None",
      timestamp: "chunk",
      batch_size: 64,
      diarise_audio: false
    }
  }
);

console.log(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 vaibhavs10/incredibly-fast-whisper using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "vaibhavs10/incredibly-fast-whisper:3ab86df6c8f54c11309d4d1f930ac292bad43ace52d10c80d87eb258b3c9f79c",
    input={
        "task": "transcribe",
        "audio": "https://replicate.delivery/pbxt/Js2Fgx9MSOCzdTnzHQLJXj7abLp3JLIG3iqdsYXV24tHIdk8/OSR_uk_000_0050_8k.wav",
        "language": "None",
        "timestamp": "chunk",
        "batch_size": 64,
        "diarise_audio": False
    }
)
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 vaibhavs10/incredibly-fast-whisper 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": "vaibhavs10/incredibly-fast-whisper:3ab86df6c8f54c11309d4d1f930ac292bad43ace52d10c80d87eb258b3c9f79c",
    "input": {
      "task": "transcribe",
      "audio": "https://replicate.delivery/pbxt/Js2Fgx9MSOCzdTnzHQLJXj7abLp3JLIG3iqdsYXV24tHIdk8/OSR_uk_000_0050_8k.wav",
      "language": "None",
      "timestamp": "chunk",
      "batch_size": 64,
      "diarise_audio": false
    }
  }' \
  https://api.replicate.com/v1/predictions

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

Output

{ "text": " the little tales they tell are false the door was barred locked and bolted as well ripe pears are fit hours fly by much too soon. The room was crowded with a mild wab. The room was crowded with a wild mob. This strong arm shall shield your honour. She blushed when he gave her a white orchid The beetle droned in the hot June sun", "chunks": [ { "text": " the little tales they tell are false the door was barred locked and bolted as well ripe pears are fit hours fly by much too soon. The room was crowded", "timestamp": [ 0, 29.72 ] }, { "text": " with a mild wab. The room was crowded with a wild mob. This strong arm shall shield your", "timestamp": [ 29.72, 38.98 ] }, { "text": " honour. She blushed when he gave her a white orchid The beetle droned in the hot June sun", "timestamp": [ 38.98, 48.52 ] } ] }

Generated in

2.8 seconds

Tweak it Report

This output was created using a different version of the model, vaibhavs10/incredibly-fast-whisper:37dfc0d6.

Run time and cost

This model costs approximately $0.0033 to run on Replicate, or 303 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia L40S GPU hardware. Predictions typically complete within 4 seconds.

Readme

Incredibly Fast Whisper

TL;DR - Transcribe 150 minutes of audio in 100 seconds - with OpenAI’s Whisper Large v3. Blazingly fast transcription is now a reality!⚡️

Optimisation type	Time to Transcribe (150 mins of Audio)
Transformers (`fp32`)	~31 (31 min 1 sec)
Transformers (`fp16` + `batching [24]` + `bettertransformer`)	~5 (5 min 2 sec)
Transformers (`fp16` + `batching [24]` + `Flash Attention 2`)	*~2 (1 min 38 sec)*
distil-whisper (`fp16` + `batching [24]` + `bettertransformer`)	~3 (3 min 16 sec)
distil-whisper (`fp16` + `batching [24]` + `Flash Attention 2`)	*~1 (1 min 18 sec)*
Faster Whisper (`fp16` + `beam_size [1]`)	~9.23 (9 min 23 sec)
Faster Whisper (`8-bit` + `beam_size [1]`)	~8 (8 min 15 sec)