CapSpeech TTS

CapSpeech TTS

A text-to-speech demo built for styled speech, where you shape the voice, accent and emotion yourself. For people who want to hear controllable speech from the OpenSound team.

Made by OpenSound 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: CapSpeech TTS is a text-to-speech demo for styled speech. Its short tagline says you can design the voice, the accent and the emotion your own way. The Space names a dataset called CapSpeech and six supporting models from other teams.

Who it suits: It suits people who want to hear how far a text-to-speech model can be steered by style.

Title: Who might use this demo
Industry | Who uses it | How they use it | Example
Speech research | Researchers | Try styled speech built around the CapSpeech dataset | Compare two emotions on one sentence
Software | Developers | Hear what controllable speech offers before adding it to a product | Test an accent on a line a product would say
Audio production | Creators | Draft a narration line in a chosen style | Generate one sentence in a chosen emotion

If you plan to use the audio commercially, you can skip it, because its license is non-commercial.

Where to use it: You can use it online or run a copy yourself.

• Web browser: the Space's page on Hugging Face, with nothing to install
• Windows, macOS or Linux: a local copy that runs with Python
• Hugging Face Spaces: where the hosted demo runs, on the zero-a10g hardware tier

Title: What the demo is good for
Use case | Who | What they get back
Hear a styled voice | Anyone curious | Speech in the voice, accent and emotion you choose
Compare settings | Researchers | Several clips of one sentence in different styles
Test before integrating | Developers | A feel for the controls a styled speech model offers

How to set it up: The hosted demo needs nothing. A local copy needs Python and pip.

• Python, with pip
• The Python packages that requirements.txt lists

1. Install the huggingface_hub package.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Install Gradio 5.31.0.
5. Install the packages from requirements.txt.
6. Start the app with Python.

pip install -U huggingface_hub
hf download OpenSound/CapSpeech-TTS --repo-type space --local-dir CapSpeech-TTS
cd CapSpeech-TTS
pip install gradio==5.31.0
pip install -r requirements.txt
python app.py

How to use it: Open the Space online, or start your copy with the command below. Then use the controls on the page to choose the voice, accent and emotion, type your text and generate.

python app.py

Title: The parts the Space names
Tool | What it does
app.py | Starts the Gradio page when you run it with Python
distil-whisper/distil-large-v2 | A speech recognition model the Space lists
google/flan-t5-large | A text model the Space lists
laion/larger_clap_music_and_speech | An audio and text matching model the Space lists
microsoft/wavlm-base-plus | A speech representation model the Space lists
microsoft/wavlm-large | A larger speech representation model the Space lists
nvidia/bigvgan_v2_24khz_100band_256x | A vocoder model the Space lists, which turns spectrograms into audio

Example: You type one sentence, pick an accent and an emotion, and press generate. The page plays that sentence back in the style you picked, and you can change one setting and generate again to hear the difference.

For newcomers and experts: If you are new to this, start with one short sentence and change a single setting at a time. That way you can tell which control changed the sound.

If you know the field, the Space pulls together a dataset named OpenSound/CapSpeech and a group of pretrained models for speech recognition, text, audio matching and vocoding. It runs Gradio 5.31.0 on zero-a10g hardware, and the capspeech folder holds the project code.

Choose a voice, accent and emotion
Type the text you want spoken
The Space's models build the speech
You hear the audio and try again

Good to know: The Space's notes are brief, so open the page to see the exact controls.

• Hosting: the hosted copy runs on the zero-a10g hardware tier
• Size: the repository holds 522 files, so the download is not tiny
• License: cc-by-nc-4
• Cost to run: none named
• Data it reaches: the text you type, plus the named models and the Python packages that get pulled in
• Not for: commercial use, since the license is non-commercial

On Hugging Face, 13 November 2026

• Likes: 101
• Built with: gradio
• License: cc-by-nc-4.0

Links

• Space: https://huggingface.co/spaces/OpenSound/CapSpeech-TTS
• Files: https://huggingface.co/spaces/OpenSound/CapSpeech-TTS/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.