Faster Whisper Large V3 Turbo Ct2

🚀 Whisper large-v3 turbo model for CTranslate2

This project offers a conversion of the deepdml/whisper-large-v3-turbo model to the CTranslate2 format. It enables seamless integration with CTranslate2 and related projects, such as faster-whisper, enhancing the efficiency of automatic speech recognition tasks.

🚀 Quick Start

Prerequisites

• Ensure you have faster-whisper and CTranslate2 installed in your Python environment.

Installation

You can install the necessary libraries using pip:

pip install faster-whisper ctranslate2

✨ Features

• Multi - language Support: Supports a wide range of languages including English, Chinese, German, Spanish, and many others.
• Efficient Conversion: Converted to the CTranslate2 format for faster inference.
• Easy Integration: Can be easily integrated into projects based on CTranslate2.

📦 Installation

The installation mainly involves getting the required Python libraries. As mentioned above, you can use pip to install faster-whisper and CTranslate2.

💻 Usage Examples

Basic Usage

from faster_whisper import WhisperModel

model = WhisperModel("deepdml/faster-whisper-large-v3-turbo-ct2")

segments, info = model.transcribe("audio.mp3")
for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

Advanced Usage

You can adjust the compute_type option when loading the model to change the quantization type. For example, if you want to use int8 quantization for more memory - efficient inference:

from faster_whisper import WhisperModel

model = WhisperModel("deepdml/faster-whisper-large-v3-turbo-ct2", compute_type="int8")

segments, info = model.transcribe("audio.mp3")
for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

📚 Documentation

Conversion details

The original model was converted with the following command:

ct2-transformers-converter --model deepdml/whisper-large-v3-turbo --output_dir faster-whisper-large-v3-turbo \
    --copy_files tokenizer.json preprocessor_config.json --quantization float16

Note that the model weights are saved in FP16. This type can be changed when the model is loaded using the compute_type option in CTranslate2.

More information

For more information about the original model, see its model card.

📄 License

This project is licensed under the MIT license.

Information Table