A Kitten TTS alternative: Supertonic 3 on an ordinary CPU
Kitten TTS from KittenML is one of the most popular open text to speech projects, released under Apache 2.0. It has two families: tiny ONNX models of 15 to 80 million parameters with eight voices that run on a CPU, and Kitten-TTS 2, a 1.7-billion-parameter 1-bit model with 115 voices, voice cloning from 5 to 30 seconds of audio, emotion tags and 20 languages.
SuperClonica runs Supertonic 3, an open model of about 99 million parameters, in 31 languages on an ordinary processor. This guide compares them honestly; we have not benchmarked Kitten TTS, so there are no speed or quality scores for it here, only facts from its repository.
Where Kitten TTS is strong
Kitten-TTS 2 clones a voice from a short clip without fine-tuning, controls expression with tags and speaks Chinese, Swahili and Hebrew, which Supertonic 3 does not. The nano model is only 15 million parameters, 25 MB in int8, small enough for very weak devices. KittenML also offers a free Kitten-TTS 2 API.
What each one needs
By its README, Kitten-TTS 2 wants a CUDA GPU with about 8 GB free or a CPU with about 6 GB of RAM, and about 2 GB of disk. Supertonic 3 is about 99 million parameters; we estimate the SuperClonica engine at 6.4 to 9.7 times faster than real time on a desktop 8-core CPU. Install it with python -m pip install superclonica-cpu and run superclonica --text "Hello" --voice F1 --lang en --output hello.mp3.
Languages and voices
Supertonic 3 speaks 31 languages, including Polish, Czech, Ukrainian, Vietnamese and Indonesian, and every voice speaks all of them. SuperClonica adds about 190 catalog voices, from narrators and announcers to game and anime characters, each a small file bought once; the ten stock voices are free and may be used commercially. Any voice page plays a line of 30 to 100 characters free.
Which to choose
Choose Kitten TTS for voice cloning from a clip, expression tags, Chinese, or the smallest possible model. Choose SuperClonica for a light CPU model in 31 languages with ready character voices, voices you train in the studio, and a speech API with a free tier of 3 minutes a day.