English TTS in just 1.4M parameters, vocoder included. PiVo-1.4M runs offline on a CPU with NumPy and ONNX Runtime, without PyTorch. 🤖 🪶 1.4M parameters, vocoder included: Generates 16 kHz mono speech with a compact acoustic model and FARGAN-derived vocoder. 🖥️ Built for local, offline use: Generate with either English voice, Pi or Vo, directly on CPU and begin playback in 10 ms audio frames after text encoding. 🗣️ Words come through clearly: PiVo reaches 0.532% WER on its held-out test set. The authors note that this measures intelligibility, not general English quality or naturalness. 🔤 Predictable English pronunciation: It expands numbers, currencies, ordinals, and ISO dates, while unsupported words raise a clear error instead of silently guessing. 📜 Code and acoustic weights are Apache-2.0. The FARGAN-derived vocoder uses Xiph/Opus BSD terms, and the pronunciation dictionary retains its CMU license.

English It’s 4:58 on Friday. Your agent has the reply ready, but volunteering your Saturday still needs your say-so. That’s the handoff in this little film. Grab Reply Desk from our free ECHO library and give your agent a clearer job before you hand over the Send button. Español Viernes, 16:58. Tu agente ya tiene la respuesta preparada, pero eso de comprometer tu sábado todavía lo decides tú. Esa es la idea de este corto. Prueba Reply Desk en nuestra biblioteca gratuita de ECHO y deja claro el trabajo del agente antes de darle acceso al botón de enviar. Tiếng Việt 4 giờ 58 chiều thứ Sáu. Trợ lý AI đã soạn xong câu trả lời, nhưng nhận việc vào thứ Bảy thì vẫn phải hỏi bạn đã. Bộ phim nhỏ này nói đúng chuyện đó. Thử thẻ Reply Desk trong thư viện ECHO miễn phí để giao việc rõ ràng trước khi cấp quyền gửi. 日本語 金曜の16時58分。AIエージェントは返信の下書きを仕上げた。でも、土曜も働く約束まで勝手にしてほしくないですよね。この短編で描いたのは、そんな役割分担です。送信権限を渡す前に、任せる仕事をはっきりさせる。無料のECHOライブラリで「Reply Desk」を試してみてください。

English ➡️ Spanish ➡️ Hindi ➡️ Mandarin Gemini 3.5 Transcribe can accurately turn speech into text across 85+ languages for building intelligent audio experiences. Watch the model seamlessly switch across 4 different languages while accurately transcribing complex numbers: