Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Edited README #2508

Open
wants to merge 9 commits into
base: main
Choose a base branch
from
Open

Edited README #2508

Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
1700765
Update README.md
kalarp Jan 4, 2025
b43abb8
Update README.md
kalarp Jan 4, 2025
eda5f55
Update README.md
kalarp Jan 4, 2025
8bca0f7
Update README.md
kalarp Jan 4, 2025
ebcf45b
Update README.md
kalarp Jan 4, 2025
81f2f2d
Update README.md
kalarp Jan 4, 2025
f2a5241
Merge pull request #1 from kalarp/README_edit
kalarp Jan 4, 2025
a2905a0
Improve structure
martab0 Jan 11, 2025
6d2c1f1
Merge pull request #2 from KomunikacjaTechnicznaVistula/main
kalarp Jan 15, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
118 changes: 79 additions & 39 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,65 +1,105 @@
# Whisper

# Whisper <!-- omit in toc -->

[[Blog]](https://openai.com/blog/whisper)
[[Paper]](https://arxiv.org/abs/2212.04356)
[[Model card]](https://github.com/openai/whisper/blob/main/model-card.md)
[[Colab example]](https://colab.research.google.com/github/openai/whisper/blob/master/notebooks/LibriSpeech.ipynb)

Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification.
## Contents <!-- omit in toc -->

- [What is Whisper](#what-is-whisper)
- [Setup](#setup)
- [Prerequsites](#prerequisites)
- [Installation](#installation)
- [Installation troubleshooting](#installation-troubleshooting)
- [Available models and languages](#available-models-and-languages)
- [Performance](#performance)
- [Command-line usage](#command-line-usage)
- [Python usage](#python-usage)
- [More examples](#more-examples)
- [License](#license)

## What is Whisper

## Approach
Whisper is a multilingual speech recognition model for general purposes, including speech translation and language identification. Whisper is trained on a large dataset of diverse audio.

![Approach](https://raw.githubusercontent.com/openai/whisper/main/approach.png)

A Transformer sequence-to-sequence model is trained on various speech processing tasks, including multilingual speech recognition, speech translation, spoken language identification, and voice activity detection. These tasks are jointly represented as a sequence of tokens to be predicted by the decoder, allowing a single model to replace many stages of a traditional speech-processing pipeline. The multitask training format uses a set of special tokens that serve as task specifiers or classification targets.
A Transformer sequence-to-sequence model is trained on various speech processing tasks. The tasks include multilingual speech recognition, speech translation, spoken language identification, and voice activity detection. These tasks are jointly represented as a sequence of tokens to be predicted by the decoder. As a result, a single model replaces many steps in traditional speech processing. The multitask training format uses a set of special tokens that serve as task specifiers or classification targets.

We used Python 3.9.9 and [PyTorch](https://pytorch.org/) 1.10.1 to train and test our models. The codebase should be compatible with Python 3.8-3.11 and recent PyTorch versions. The codebase also depends on a few Python packages, most notably [OpenAI's tiktoken](https://github.com/openai/tiktoken) for their fast tokenizer implementation.

## Setup

We used Python 3.9.9 and [PyTorch](https://pytorch.org/) 1.10.1 to train and test our models, but the codebase is expected to be compatible with Python 3.8-3.11 and recent PyTorch versions. The codebase also depends on a few Python packages, most notably [OpenAI's tiktoken](https://github.com/openai/tiktoken) for their fast tokenizer implementation. You can download and install (or update to) the latest release of Whisper with the following command:

pip install -U openai-whisper

Alternatively, the following command will pull and install the latest commit from this repository, along with its Python dependencies:

pip install git+https://github.com/openai/whisper.git
### Prerequisites

To update the package to the latest version of this repository, please run:

pip install --upgrade --no-deps --force-reinstall git+https://github.com/openai/whisper.git

It also requires the command-line tool [`ffmpeg`](https://ffmpeg.org/) to be installed on your system, which is available from most package managers:
* Whisper requires the command-line tool [`ffmpeg`](https://ffmpeg.org/) to be installed on your system. The command-line tool is available from most package managers. To install [`ffmpeg`](https://ffmpeg.org/), use one of the following commands for your operating system:

**on Ubuntu or Debian**
```bash
# on Ubuntu or Debian
sudo apt update && sudo apt install ffmpeg
```

# on Arch Linux
**on Arch Linux**
```bash
sudo pacman -S ffmpeg
```

# on MacOS using Homebrew (https://brew.sh/)
**on MacOS using Homebrew (https://brew.sh/)**
```bash
brew install ffmpeg
```

# on Windows using Chocolatey (https://chocolatey.org/)
**on Windows using Chocolatey (https://chocolatey.org/)**
```bash
choco install ffmpeg
```

# on Windows using Scoop (https://scoop.sh/)
**on Windows using Scoop (https://scoop.sh/)**
```bash
scoop install ffmpeg
```

You may need [`rust`](http://rust-lang.org) installed as well, in case [tiktoken](https://github.com/openai/tiktoken) does not provide a pre-built wheel for your platform. If you see installation errors during the `pip install` command above, please follow the [Getting started page](https://www.rust-lang.org/learn/get-started) to install Rust development environment. Additionally, you may need to configure the `PATH` environment variable, e.g. `export PATH="$HOME/.cargo/bin:$PATH"`. If the installation fails with `No module named 'setuptools_rust'`, you need to install `setuptools_rust`, e.g. by running:
* If [tiktoken](https://github.com/openai/tiktoken) does not provide a pre-built wheel for your platform, install [`rust`](http://rust-lang.org). Follow the [Getting started page](https://www.rust-lang.org/learn/get-started) to install the Rust development environment.

### Installation

* You can download and install (or update to) the latest release of Whisper with the following command:

```bash
pip install setuptools-rust
pip install -U openai-whisper
```

* Alternatively, use the following command to pull and install the latest commit from this repository and its Python dependencies:

```bash
pip install git+https://github.com/openai/whisper.git
```

* To update the package to the latest version of this repository, run:

```bash
pip install --upgrade --no-deps --force-reinstall git+https://github.com/openai/whisper.git
```

### Installation troubleshooting

If you see installation errors during the installation of Whisper, follow these steps:
* Check if you have [`rust`](http://rust-lang.org) installed on your system. If not, follow the [Getting started page](https://www.rust-lang.org/learn/get-started) to install the Rust development environment.
* Additionally, you may need to configure the `PATH` environment variable. You can use the following command:
```bash
export PATH="$HOME/.cargo/bin:$PATH"
```
* If the installation fails with `No module named 'setuptools_rust'`, install `setuptools_rust`. You can use the following command:

```bash
pip install setuptools-rust
```

## Available models and languages

There are six model sizes, four with English-only versions, offering speed and accuracy tradeoffs.
Below are the names of the available models and their approximate memory requirements and inference speed relative to the large model.
The relative speeds below are measured by transcribing English speech on a A100, and the real-world speed may vary significantly depending on many factors including the language, the speaking speed, and the available hardware.
There are six model sizes, four with English-only versions, offering a compromise between speed and accuracy. In the table below are the names of the available models, their approximate memory requirements and their inference speed relative to the large model. The relative speeds given in the table are measured by transcribing English speech on the A100 graphics processing unit (GPU). The real-world speed may vary significantly depending on many factors including the language, the speaking speed, and the available hardware.

| Size | Parameters | English-only model | Multilingual model | Required VRAM | Relative speed |
|:------:|:----------:|:------------------:|:------------------:|:-------------:|:--------------:|
Expand All @@ -71,35 +111,36 @@ The relative speeds below are measured by transcribing English speech on a A100,
| turbo | 809 M | N/A | `turbo` | ~6 GB | ~8x |

The `.en` models for English-only applications tend to perform better, especially for the `tiny.en` and `base.en` models. We observed that the difference becomes less significant for the `small.en` and `medium.en` models.
Additionally, the `turbo` model is an optimized version of `large-v3` that offers faster transcription speed with a minimal degradation in accuracy.
Additionally, the `turbo` model is an optimized version of `large-v3`. It offers faster transcription speed with a minimal degradation in accuracy.

Whisper's performance varies widely depending on the language. The figure below shows a performance breakdown of `large-v3` and `large-v2` models by language, using WERs (word error rates) or CER (character error rates, shown in *Italic*) evaluated on the Common Voice 15 and Fleurs datasets. Additional WER/CER metrics corresponding to the other models and datasets can be found in Appendix D.1, D.2, and D.4 of [the paper](https://arxiv.org/abs/2212.04356), as well as the BLEU (Bilingual Evaluation Understudy) scores for translation in Appendix D.3.

![WER breakdown by language](https://github.com/openai/whisper/assets/266841/f4619d66-1058-4005-8f67-a9d811b77c62)
## Performance

Whisper's performance varies widely by language. The figure below shows a performance breakdown of `large-v3` and `large-v2` models by language. The performance breakdown uses Word Error Rates (WER) or Character Error Rates (CER, shown in *Italics*) evaluated on the Common Voice 15 and Fleurs datasets. Additional Word Error Rates or Character Error Rates metrics corresponding to the other models and datasets can be found in:
* Appendix D.1, D.2, and D.4 of [the paper](https://arxiv.org/abs/2212.04356).
* The Bilingual Evaluation Understudy (BLEU) scores for translation in Appendix D.3.

![WER breakdown by language](https://github.com/openai/whisper/assets/266841/f4619d66-1058-4005-8f67-a9d811b77c62)

## Command-line usage

The following command will transcribe speech in audio files, using the `turbo` model:
The following command will transcribe speech in audio files. The command uses the `turbo` model:

whisper audio.flac audio.mp3 audio.wav --model turbo

The default setting (which selects the `turbo` model) works well for transcribing English. To transcribe an audio file containing non-English speech, you can specify the language using the `--language` option:

whisper japanese.wav --language Japanese

Adding `--task translate` will translate the speech into English:
Add `--task translate` to translate the speech into English:

whisper japanese.wav --language Japanese --task translate

Run the following to view all available options:
Run the following command to view all available options:

whisper --help

See [tokenizer.py](https://github.com/openai/whisper/blob/main/whisper/tokenizer.py) for the list of all available languages.


## Python usage

Transcription can also be performed within Python:
Expand All @@ -112,9 +153,9 @@ result = model.transcribe("audio.mp3")
print(result["text"])
```

Internally, the `transcribe()` method reads the entire file and processes the audio with a sliding 30-second window, performing autoregressive sequence-to-sequence predictions on each window.
Internally, the `transcribe()` method reads the entire file and processes the audio with a sliding 30-second window. The method performs autoregressive sequence-to-sequence predictions on each window.

Below is an example usage of `whisper.detect_language()` and `whisper.decode()` which provide lower-level access to the model.
Below is an example usage of `whisper.detect_language()` and `whisper.decode()` which provide lower-level access to the model:

```python
import whisper
Expand Down Expand Up @@ -142,9 +183,8 @@ print(result.text)

## More examples

Please use the [🙌 Show and tell](https://github.com/openai/whisper/discussions/categories/show-and-tell) category in Discussions for sharing more example usages of Whisper and third-party extensions such as web demos, integrations with other tools, ports for different platforms, etc.

Use the [🙌 Show and tell](https://github.com/openai/whisper/discussions/categories/show-and-tell) category in Discussions for sharing more example usages of Whisper and third-party extensions such as web demos, integrations with other tools or ports for different platforms.

## License

Whisper's code and model weights are released under the MIT License. See [LICENSE](https://github.com/openai/whisper/blob/main/LICENSE) for further details.
Whisper's code and model weights are released under the Massachusetts Institute of Technology (MIT) License. See [LICENSE](https://github.com/openai/whisper/blob/main/LICENSE) for further details.