Published on September 17, 2025, the announcement describes an open text-to-speech model supporting long conversational speech, multilingual and cross-lingual voice cloning, low latency, and random timbre generation.
On September 17, 2025, FireRedTTS 2 was announced as an open text-to-speech model. The publication highlights long conversational speech, multilingual and cross-lingual voice cloning, low latency, and random timbre generation. It also says engineers can evaluate a runnable model independently for multilingual speech generation and voice cloning.
To consult and verify the details, find the original model publication and check its description, examples, and terms of use; the brief announcement provides no quantitative results. If using AI to study or test the material, avoid submitting identifiable recordings or personal data without authorization, and follow your organization’s policies. Rota Nacional’s voice features are in preview, so they are not an available capability for this test.