faster-whisper-small

faster-whisper-small

Whisper small converted to the CTranslate2 format so faster-whisper can transcribe audio with it. For developers who want local speech-to-text in Python.

Made by Systran on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This repository holds OpenAI's Whisper small speech recognition model, converted to the CTranslate2 format. It works in CTranslate2 and in projects built on it, such as faster-whisper. The conversion keeps the weights in FP16 by default.

Who it suits: Developers who want to turn audio into text with a small Whisper model on their own computer, through Python.

Industry | Who uses it | How they use it | Example
Media | Podcasters and video editors | Transcribe recordings into text | Turn an interview file into a transcript
Software | Developers adding speech to text | Call the model from Python code | Print timed segments from an audio file
Education | Teachers and students | Transcribe lectures and talks | Convert a recorded lecture to text

You can skip it if you need the original model files for another framework, since this copy targets CTranslate2.

Where to use it:

• Python, with the faster-whisper package
• Any CTranslate2-based project
• Windows, macOS or Linux, wherever Python runs

Use case | Who | What they get back
Transcribe an audio file | Developers | Text segments with start and end times
Build a transcription tool | App makers | Speech recognition inside your own program
Run speech recognition locally | Privacy-minded users | Transcripts made without sending audio elsewhere

How to set it up: You need Python and the faster-whisper package. The model name small makes the library fetch these files by itself, or you can download them first.

• Python and pip
• The faster-whisper package
• An audio file to transcribe

1. Install the faster-whisper package.
2. Optionally download the files with the Hugging Face tool.
3. Load the model by the name small.

pip install -U huggingface_hub
hf download Systran/faster-whisper-small

How to use it: Load the model, call transcribe on an audio file, and loop over the segments it returns. Each segment carries start and end times and its text.

from faster_whisper import WhisperModel

model = WhisperModel("small")

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

Tool | What it does
WhisperModel | Loads the model by name
transcribe | Returns the segments and information about the audio
segment.start, segment.end, segment.text | Give the timing and the text of each segment

Example: You point transcribe at a file named audio.mp3. The loop prints one line per segment, with its start and end in seconds followed by the recognized words.

For newcomers and experts: If you are new, run the snippet above on a short recording and read the printed lines. Larger models cost more time and memory, so small is a gentle first choice.

For experts, the conversion was done with the CTranslate2 transformers converter at float16, copying the tokenizer file. You can change the precision when you load the model with the compute_type option. The card sends you to the original Whisper small page for details on the model itself.

Audio file
faster-whisper loads the CTranslate2 model
Segments with timestamps
Text you print or save

Good to know:

• Conversion: made with ct2-transformers-converter at float16
• Precision: weights are saved in FP16, and compute_type changes it at load time
• Languages: the record lists English, Chinese, German, Spanish, Russian, Korean, French and Japanese
• License: MIT
• Cost to run: no API key or paid service named
• Data it reaches: downloads files from Hugging Face, and the audio you transcribe stays on your machine
• Not for: use outside CTranslate2, since this is a converted copy

On Hugging Face, 13 November 2026

• Task: automatic speech recognition
• Library: ctranslate2
• Downloads in the last 30 days: 5,890,728
• Likes: 95
• License: mit
• Languages: en, zh, de, es, ru, ko, fr, ja
• Fine-tunes: 1
• Tags: ctranslate2, audio, automatic-speech-recognition

Links

• Model: https://huggingface.co/Systran/faster-whisper-small
• Files: https://huggingface.co/Systran/faster-whisper-small/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Systran/faster-whisper-small

Then follow the usage in its model card. Nothing in this kit runs by itself.