{"name":"tiny-voca","repo":"n4ze3m/tiny-voca","url":"https://midorreal.com/project/n4ze3m-tiny-voca","repository":"https://github.com/n4ze3m/tiny-voca","description":"A 4.5M-parameter two-voice English text-to-speech model, trained from scratch","facts":{"license":{"value":"Apache-2.0","source":"https://github.com/n4ze3m/tiny-voca","checked":"22 Nov 2026"},"language":{"value":"Python","source":"https://github.com/n4ze3m/tiny-voca","checked":"22 Nov 2026"},"homepage":{"value":"https://tiny-voca.n4ze3m.com","source":"https://tiny-voca.n4ze3m.com","checked":"22 Nov 2026"},"releases":{"value":"0","source":"https://github.com/n4ze3m/tiny-voca/releases","checked":"19 Nov 2026"},"contributors":{"value":"2","source":"https://github.com/n4ze3m/tiny-voca/graphs/contributors","checked":"19 Nov 2026"},"archived":{"value":"no","source":"https://github.com/n4ze3m/tiny-voca","checked":"22 Nov 2026"}},"stars":828.0,"writeup":{"what_it_is":"tiny-voca is a small English text-to-speech model with 4.5M parameters and two voices, F2 (female) and M2 (male). The model was trained from scratch and the project includes code, weights, a training recipe and a dataset.","audience":"Developers who want a small, local, open English text-to-speech model.","claims":[{"kind":"specific","claim":"The model has 4,537,291 parameters and is 18 MB in size.","excerpt":"4,537,291 parameters, 18 MB, waveform generator included.","status":"not_checked"},{"kind":"specific","claim":"It runs at about 70x real time on four CPU threads.","excerpt":"about 70x real time on four CPU threads.","status":"not_checked"},{"kind":"specific","claim":"It has two American English voices, F2 and M2, at 24 kHz.","excerpt":"F2 (female) and M2 (male), American English, 24 kHz.","status":"not_checked"},{"kind":"specific","claim":"It runs under PyTorch, ONNX Runtime, and ONNX Runtime Web.","excerpt":"PyTorch, ONNX Runtime, and ONNX Runtime Web.","status":"not_checked"},{"kind":"specific","claim":"The project is open, including code, weights, training recipe and dataset.","excerpt":"code, weights, training recipe and dataset. Trained from scratch.","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-11-16"},"why_now":null,"owner_supplied":null,"score":{"verdict":"sleeper","hype":42,"reality":81,"gap":-39,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":null,"changes":[{"date":"2026-11-14T16:02:56+00:00","text":"Repository created","source":"https://github.com/n4ze3m/tiny-voca"}],"cite":"tiny-voca is a GitHub project at https://github.com/n4ze3m/tiny-voca written mainly in Python, described there as \"A 4.5M-parameter two-voice English text-to-speech model, trained from scratch\".","attribution":"Data from Mid or Real, https://midorreal.com/project/n4ze3m-tiny-voca, checked 22 November 2026. Attribution with a link is required.","still_real":null}