Best open-source speech models for transcription and text-to-speech

Speech models turn audio into text and text into audio, and most of them are small enough for a laptop. This page covers both directions. Our dedicated lists for text-to-speech and speech recognition go deeper on each direction. Always spot-check transcripts and be careful with voice cloning: using someone's voice without permission can be illegal.

  • List updated November 11, 2026
  1. Nemotron 3.5 Transcribe

    NVIDIA's speech-to-text model with 40 languages, speaker labels, translation and prompt-controlled transcript style, under the permissive OpenMDW license. It processes finished recordings rather than live streams.

    • NVIDIA
    • 34B (4.3B active)
    • 64K tokens
    • OpenMDW-1.1 (Permissive)
    • from 25 GB
  2. Whisper large-v3-turbo

    OpenAI's MIT-licensed speech recognition model. The turbo version is about 8x faster than large and needs roughly 6 GB of GPU memory. It can hallucinate text that was never spoken, so check important transcripts.

    • OpenAI
    • 1.54B
    • MIT (Permissive)
    • from 6 GB
  3. Qwen3-ASR

    Alibaba's Apache 2.0 speech-to-text pair (1.7B and 0.6B): 30 languages plus 22 Chinese dialects, singing, streaming and timestamps.

    • Alibaba Qwen
    • 2.35B
    • Apache 2.0 (Permissive)
    • from 2.12 GB
  4. Cohere Transcribe

    Cohere's Apache 2.0 2B model that only transcribes: 14 languages, with no timestamps, no speaker labels and no automatic language detection.

    • Cohere
    • 2.07B
    • Apache 2.0 (Permissive)
    • from 4.75 GB
  5. Voxtral Realtime

    Mistral's Apache 2.0 4B streaming speech-to-text model: 13 languages and a delay you set between 80 ms and 2.4 s, for live use.

    • Mistral AI
    • 4.43B
    • 128K tokens
    • Apache 2.0 (Permissive)
    • from 10 GB
  6. Confucius4-R2T2

    NetEase Youdao's streaming speech-to-text model for Chinese and English, adapted from Qwen3-ASR-1.7B. It never revises committed words, and the license needs a separate agreement above 100 million monthly users or RMB 1 billion of revenue.

    • NetEase Youdao
    • 2.04B
    • NetEase Youdao Model Use (Conditional)
    • from 4.69 GB
  7. Qwen3-TTS

    Alibaba's Apache 2.0 text-to-speech family in 1.7B and 0.6B sizes: 10 languages, voice cloning from about 3 seconds of audio, voice design from a description, and streaming latency as low as 97 ms.

    • Alibaba Qwen
    • 1.92B
    • Apache 2.0 (Permissive)
    • from 3.68 GB
  8. Kokoro Most downloaded

    hexgrad's Apache 2.0 82M text-to-speech model with 54 preset voices in 8 languages. It is the most downloaded speech model here, with about 12.5 million downloads in the 30 days to 11 November 2026, but it cannot clone voices.

    • hexgrad
    • 82M
    • Apache 2.0 (Permissive)
    • from 0.38 GB
  9. VoxCPM2

    OpenBMB's 2B Apache 2.0 text-to-speech model: 30 languages, voice design from a text description, cloning and 48 kHz output on about 8 GB of VRAM.

    • OpenBMB
    • 2.29B
    • Apache 2.0 (Permissive)
    • from 8 GB
  10. IndexTTS-2.5

    Bilibili's 0.8B zero-shot cloner for Chinese, English, Japanese, Spanish and Arabic, with separate emotion, pronunciation and speed controls on about 6 GB of VRAM. Its own license has no general commercial ban but sets 100 million-user and RMB 1 billion thresholds.

    • Bilibili
    • 800M
    • Bilibili Model Use (Conditional)
    • from 6 GB
  11. AuK

    Tencent's MIT-licensed 1.5B model that handles speech generation, editing, enhancement and separation as written instructions. Peak memory is about 25 GiB in bf16, or about 17 GiB with CPU offload.

    • Tencent
    • 1.5B
    • MIT (Permissive)
    • from 18 GB
  12. ZONOS3-Mini

    Zyphra's Apache 2.0 800M text-to-speech model with voice cloning in 43 languages, released on 9 November 2026. Zyphra says it runs in real time on laptops, phones and in a browser.

    • Zyphra
    • 800M (100M active)
    • Apache 2.0 (Permissive)
    • from 10 GB
  13. Pocket TTS

    Kyutai's 100M text-to-speech model with voice cloning that runs faster than real time on a laptop CPU. The weights are CC BY 4.0 behind a prohibited-use gate.

    • Kyutai
    • 100M
    • CC BY 4.0 (Permissive)
    • from 0.127 GB
  14. Drama 3

    Fish Audio's 36B text-to-speech mixture of experts with 83 languages and performance directions written into the text. The license is free only below US$500,000 of yearly revenue, and it never covers a hosted voice service.

    • Fish Audio
    • 35.9B (3B active)
    • Fish Audio Community 1.0 (Conditional)
    • from 25 GB
  15. OmniVoice

    Xiaomi's text-to-speech model covers 600+ languages with cloning and voice design. The code is Apache 2.0, but the weights are CC BY-NC, so it is for non-commercial use.

    • Xiaomi
    • 613M
    • CC BY-NC (Non-commercial)
    • from 3.75 GB
  16. LFM3-Audio-2B

    Liquid AI's 2B on-device voice model: speech in and out, transcription, synthesis and translation across 18 languages. The license is free only below US$10 million of revenue.

    • Liquid AI
    • 2B
    • 32K tokens
    • LFM Open 1.0 (Conditional)
    • from 4.6 GB
  17. Sophon-Realtime

    Adventists AI's open full-duplex voice model, released on 8 November 2026 under Apache 2.0: a 32B thinker that listens while you talk, handles interruptions and can laugh or sigh, with three builds and 0.25 to 0.6 s to first audio. It needs one 80 GB GPU, and the benchmark figures are the developer's own.

    • Adventists AI
    • 32B
    • Apache 2.0 (Permissive)
    • from 80 GB

How we chose

  • Open weights for speech recognition, speech synthesis or both.
  • License terms read in full, with voice-cloning and revenue conditions noted.
  • Language counts and latency figures are the developers' own claims.