Speechify Simba 3.5: one voice model, 48 languages
Speechify launches Simba 3.5, a single text-to-speech model covering 48 languages, up from six in Simba 3.0. In blind listening tests native listeners preferred it in 62% of comparisons, and median time to first audio is 106 ms.
What the article says
Speechify introduces Simba 3.5, the latest version of its text-to-speech model family, launched on October 21, 2026. A single model now handles 48 languages, up from six in Simba 3.0, including regional variants such as Brazilian and European Portuguese. All catalog voices, plus a voice cloned on a paid plan, can read these languages; a voice keeps the accent of its original recording, however. In blind listening tests on the six languages both models share, native listeners preferred Simba 3.5 over Simba 3.0 Multilingual in 62% of comparisons.
For developers, the model is available on all SpeechifyAI plans, including the free one, with the ID simba-3.5 on the speech, streaming and WebSocket endpoints. Voice IDs, audio formats, word timings and SSML from Simba 3.2 stay the same, and a request with no explicit model still uses Simba 3.0. Until November 21, 2026, Simba 3.5 is billed at Simba 3.2 rates.
Read the official article AnnuIA profile
Summary written by AnnuIA in its own words from the official page of Speechify. The content and rights belong to the publisher: please read the original article.