Automatic Speech Recognition

Automatic Speech Recognition

A Gradio demo from the k2-fsa project that turns speech audio into text. It names six speech models, including Vosk Russian models and icefall models, and runs on CPU.

Made by k2-fsa on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This Space is an automatic speech recognition demo from the k2-fsa project. It turns speech audio into written text. The repository names six speech models, two Vosk models with ru in their names and four icefall models, and it runs on basic CPU hardware at Hugging Face.

Who it suits: People who want to try speech to text models without installing anything, and developers who want a small example of several recognizers behind one Gradio page. Speech researchers can see which named models the demo covers.

Industry | Who uses it | How they use it | Example
Speech technology | Speech engineers | Try the named recognizers on their own audio | Running one recording through the demo and reading the text
Media | Editors and journalists | Turn a spoken recording into rough text | Getting a first transcript of an interview clip
Software | Python developers | Read app.py, decode.py and model.py to see how the models are called | Copying the Space to add another model
Research | Students of speech recognition | Compare what the named models produce | Hearing how LibriSpeech-trained models handle a clip

If you never work with speech audio, you can skip it.

Where to use it:

• A web browser, on the Space's page at Hugging Face
• Windows with Python 3.10.0
• macOS with Python 3.10.0
• Linux with Python 3.10.0

Use case | Who | What they get back
Transcribe a recording | Anyone with speech audio | The text the model hears
Try several recognizers | Engineers and students | A demo that covers the six models it names

How to set it up: Online use needs nothing. A copy on your own computer needs a little.

• Python 3.10.0 on Windows, macOS or Linux
• The pip tool, to install the packages from requirements.txt
• Nothing at all if you use the Space online, since it runs in the browser

1. Install the Hugging Face hub package.
2. Download the Space's files into a local folder.
3. Move into that folder.
4. Install Gradio at the pinned version, then the packages from requirements.txt.
5. Start the app.

pip install -U huggingface_hub
hf download k2-fsa/automatic-speech-recognition --repo-type space --local-dir automatic-speech-recognition
cd automatic-speech-recognition
pip install gradio==5.25.2
pip install -r requirements.txt
python app.py

How to use it: Online, open the Space's page and use the app there. On your own computer, run this command from the project folder and open the page Gradio shows. The repository also holds a test_wavs folder of sample audio.

python app.py

Example: You open the Space and give it a speech recording. The app turns the recording into text and shows you what it heard.

For newcomers and experts: If you are new to this, open the Space's page and try one of the sample recordings, if the page offers them. Nothing needs installing, and you see what speech to text looks like.

If you know the field, the six named models are two Vosk models with ru in their names, an icefall MGB2 conformer CTC model and three icefall LibriSpeech pruned transducer models. The code is split into app.py, decode.py, model.py and examples.py, and the hosted copy runs on CPU, so long recordings may be slow.

Open the Space or run app.py locally
Give the app a speech recording
A named model decodes the audio
You read the text it returns

Good to know: The source says little beyond the model names, so check the results on your own audio.

• License: Apache 2.0, as the Space's license line says
• Cost to run: none named, and the hosted copy uses cpu-basic hardware at Hugging Face
• Data it reaches: the audio you give it, and the six speech models it names, which are hosted on Hugging Face

On Hugging Face, 13 November 2026

• Likes: 60
• Built with: gradio
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/k2-fsa/automatic-speech-recognition
• Files: https://huggingface.co/spaces/k2-fsa/automatic-speech-recognition/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.