tiny-voca is now on Pinokio: 1-click install
tiny-voca by n4ze3m is a tiny English text-to-speech model with just 4.5M parameters. It runs on the CPU at roughly
70x real time, so you don't need a GPU. The launcher installs it with a CPU build of PyTorch and runs a simple Gradio
web UI on your own machine.
๐ง What it does
๐ง One 4.5M-parameter model. The weights and phonemizer are only about 24 MB.
๐๏ธ Comes with 2 voices: F2 (female) and M2 (male).
โก Runs on the CPU at about 70x real time and outputs 24 kHz mono audio.
๐๏ธ Has 4 controls: Speed, Pitch shift (in semitones), Pitch range (more or less pitch movement) and Voice.
๐ Text of any length works. It's read sentence by sentence and joined into one file.
๐ Includes a local Gradio API at /generate for your own apps and scripts (Python, JS and curl examples are in the
README).
๐ฌ Example
Hello! This is tiny voca, a small text to speech model.
๐ Before you install
๐ป English only.
๐ฆ On the first install, the weights (about 24 MB) download automatically. After that it runs offline.
๐ Update pulls the latest tiny-voca, the dizione phonemizer and new weights.
๐ The tiny-voca code and weights are Apache 2.0: https://huggingface.co/n4ze3m/tiny-voca
๐ The launcher is MIT.
๐ Install โ Start โ Generate
https://github.com/PierrunoYT/Tiny-Voca-TTS-Pinokio
tiny-voca is now on Pinokio: 1-click install
tiny-voca by n4ze3m is a tiny English text-to-speech model with just 4.5M parameters. It runs on the CPU at roughly
70x real time, so you don't need a GPU. The launcher installs it with a CPU build of PyTorch and runs a simple Gradio
web UI on your own machine.
๐ง What it does
๐ง One 4.5M-parameter model. The weights and phonemizer are only about 24 MB.
๐๏ธ Comes with 2 voices: F2 (female) and M2 (male).
โก Runs on the CPU at about 70x real time and outputs 24 kHz mono audio.
๐๏ธ Has 4 controls: Speed, Pitch shift (in semitones), Pitch range (more or less pitch movement) and Voice.
๐ Text of any length works. It's read sentence by sentence and joined into one file.
๐ Includes a local Gradio API at /generate for your own apps and scripts (Python, JS and curl examples are in the
README).
๐ฌ Example
Hello! This is tiny voca, a small text to speech model.
๐ Before you install
๐ป English only.
๐ฆ On the first install, the weights (about 24 MB) download automatically. After that it runs offline.
๐ Update pulls the latest tiny-voca, the dizione phonemizer and new weights.
๐ The tiny-voca code and weights are Apache 2.0: https://huggingface.co/n4ze3m/tiny-voca
๐ The launcher is MIT.
๐ Install โ Start โ Generate
https://github.com/PierrunoYT/Tiny-Voca-TTS-Pinokio