tiny-voca text to speech

A 4.5 million parameter English text to speech model. Two voices, F2 and M2, 24 kHz output. It runs about 70 times faster than real time on four CPU threads, so a sentence comes back almost instantly.

Everything happens on the weights in n4ze3m/tiny-voca. The phonemizer is the author's dizione front end, shipped with the model.

Length limit: one utterance per call, the model rejects text that exceeds its token window. Split long paragraphs into sentences.

Other demos on this account: background remover, browser transcription, tiny-voca text to speech, FLUX.2 klein inpainting.

Voice
0.7 1.4
-0.12 0.12
0.5 1.5