FASTDIAR

FASTDIAR

Labels who is speaking in an audio file or live microphone stream as it plays, about one second behind. A demo for speech developers and researchers who work on speaker diarization.

Made by herimor 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: FASTDIAR is a live demo of streaming speaker diarization, which means working out who speaks when. Its approach pairs a speaker encoder that works on each audio frame with clustering done online. You can add an audio file or use your microphone.

The waveform on the page changes color for each speaker, and the labels settle roughly a second after the audio plays. When you press Stop, you can download the result as an RTTM file.

Who it suits: Speech developers and researchers who want to see streaming diarization work on their own audio.

Industry | Who uses it | How they use it | Example
Speech technology | Developers | Try live speaker labels before building on them | Speak into the microphone and watch the colors
Research | Diarization researchers | Check how labels settle as audio streams | Play a recorded file and read the output
Meetings and media | Teams with recordings | See who spoke when in a recording | Download the RTTM after pressing Stop

You can skip it if you only need transcripts, since this demo labels speakers and does not write out what they say.

Where to use it:

• A web browser, on the Space's page at Hugging Face, with nothing to install
• Windows, macOS or Linux with Python, from a copy you download

Use case | Who | What they get back
Label speakers in a file | Researchers | A waveform colored by speaker
Label speakers live | Developers | Colors that settle about a second behind the audio
Export the labels | Analysts | An RTTM file after pressing Stop

How to set it up:

• Python 3.12 on your computer
• The Python packages listed in requirements.txt
• A connection when the Space starts, since the code and checkpoint are fetched then
• The huggingface_hub package, which gives you the hf command

1. Install the huggingface_hub package.
2. Download the Space with the hf command.
3. Move into the downloaded folder.
4. Install the pinned Gradio version.
5. Install the packages from requirements.txt.
6. Start the app with Python.

pip install -U huggingface_hub
hf download herimor/fastdiar --repo-type space --local-dir fastdiar
cd fastdiar
pip install gradio==6.28.0
pip install -r requirements.txt
python app.py

How to use it: Open the Space's page in your browser, or start your own copy with the command below. Add an audio file or switch on the microphone, watch the colors appear, then press Stop to get the RTTM download.

python app.py

Example: You speak into your microphone with a friend. The waveform on the page changes color by speaker, and after you press Stop you can download the labels as an RTTM file.

For newcomers and experts: If you are new to diarization, try a short recording with two voices. The color changes show where the model thinks the speaker switched.

If you know the field, the demo runs the large model on CPU. Its code is installed from the project's GitHub repository and the checkpoint comes from that project's release when the Space starts. Output lags the audio by about a second because labels are finalized after a short delay.

You add a file or use the microphone
The speaker encoder reads each frame
Online clustering assigns a speaker
Colors appear and Stop offers the RTTM

Good to know:

• License: the Space lists mit
• Cost to run: none named
• Data it reaches: the audio you add or record, plus the code and checkpoint it fetches from the project's GitHub repository when it starts
• Not for: speech to text, since it labels speakers and does not transcribe

On Hugging Face, 13 November 2026

• Likes: 3
• Built with: gradio
• License: mit

Links

• Space: https://huggingface.co/spaces/herimor/fastdiar
• Files: https://huggingface.co/spaces/herimor/fastdiar/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.