RISC-V Sherpa-Onnx TTS on the Sipeed Lichee Console
Update: See the pinned comment to run this faster with a different model.
In an earlier video I showed Stable Diffusion on the Sipeed Lichee Console.
Now I'm testing Sherpa Onnx TTS (text to speech).
I wasn't able to compile it, but they do supply binaries.
And so far I have only been able to get the vits-ljs model working. But unfortunately it is very slow (more than two minutes for a four second audio file).
https://github.com/k2-fsa/sherpa-onnx
Example command:
cd /path/to/sherpa-onnx
./build/bin/sherpa-onnx-offline-tts \
--vits-model=./vits-ljs/vits-ljs.onnx \
--vits-lexicon=./vits-ljs/lexicon.txt \
--vits-tokens=./vits-ljs/tokens.txt \
--output-filename=./liliana.wav \
'liliana, the most beautiful and lovely assistant of our team!'
00:00 Intro
01:20 Sherpa-Onnx
04:04 Downloading Binaries
06:00 Downloading ljs Model
06:27 Example Command
If this video was helpful, please like, comment and subscribe!
Bluesky: https://bsky.app/profile/livinglinux....
#riscv #sipeed #licheepi #ai #generativeai