๐ Phonon TTS is now on Pinokio: 1-click install
Phonon is Gradium's family of small text-to-speech models (31Mโ40M parameters). They run entirely on your CPU and
don't need a GPU or PyTorch.
๐ง What it does
๐ Reads text aloud in English, French, German, Spanish and Portuguese. One multilingual model covers all five.
๐ง Includes all 7 Phonon models: phonon-31m-en plus the 40m models for en, fr, de, es, pt and multilingual.
๐๏ธ Comes with about 100 voices, filtered by language. Each voice downloads the first time you use it.
๐ข Spells out numbers, currency, dates and symbols in the selected language, so "$5" is read as "5 dollars".
๐๏ธ Has a speed slider and outputs natural 24 kHz WAV.
๐ Includes a local Gradio API at /generate for your own apps and scripts.
๐ฌ Example
The meeting moved to 3:30 PM on 12/05, and tickets now cost $5.
๐ Before you install
๐ป Runs on Windows, Linux and macOS. On x86_64 Windows and Linux the CPU needs AVX2/FMA (most CPUs from 2015 on). Each
model is only 65โ75 MB.
๐ phonon-31m-en works out of the box with no account and is Apache 2.0.
The 40m models are gated. Accept their terms on Hugging Face, then paste a read token into the launcher's ๐ Hugging
Face Token button. You don't need to restart.
huggingface.co/collections/gradโฆ
๐ The 40m models are CC BY-NC 4.0, so they're for non-commercial use only.
๐ Install โ Start โ Generate
github.com/PierrunoYT/Phonon-TTโฆ
#phonon
Browse tags
Most posts ยท Last 30 days
No recent tagged posts yet.
Youโre all caught up.
Join the community
Follow people and apps for updates. Explore Announcements and All for more.
Discover people โ
