voice & speech: 49 curated resources
voice & speech
49 resources · page 1 of 2
AssemblyAI Universal-3.5 Pro
by AssemblyAI
AssemblyAI documentation for Universal-3.5 Pro: pre-recorded speech-to-text in 18 languages with code switching and prompting.
Breeze TTS 2
by Breeze Blue
BreezeBlue announcement of Breeze TTS 2, an open-weight English and Chinese text-to-speech model with voice design, cloning and streaming output.
cohere-transcribe-03-2026
by Cohere Labs
Hugging Face model card for CohereLabs/cohere-transcribe-03-2026, a 2B-parameter open speech recognition model for 14 languages.
CosyVoice
by QwenAudio
QwenAudio CosyVoice GitHub repo: multilingual voice generation model with inference, training and deployment.
faster-whisper (Systran)
by SYSTRAN
SYSTRAN faster-whisper GitHub repo: Whisper reimplementation on CTranslate2 for faster transcription.
Fun-Realtime-ASR-preview
Artificial Analysis speech-to-text page for Alibaba, covering Fun-Realtime-ASR-preview: word error rate, speed and price compared with other models.
Gemini 2.5 Pro
Artificial Analysis speech-to-text page for Google, covering Gemini 2.5 Pro: word error rate, speed and price compared with other models.
GigaAM (Sber)
Sber GigaAM GitHub repo: foundational models for Russian speech recognition.
GigaAM-v3: открытая SOTA-модель распознавания речи на русском языке
by Sber
Sber developers article on GigaAM-v3, an open MIT-licensed Russian speech recognition model released on GitHub and Hugging Face.
Gladia Solaria-3
by Gladia
Gladia page for Solaria-3, its pre-recorded speech-to-text model for English, French, German, Spanish and Italian.
GPT Transcribe
Artificial Analysis speech-to-text page for OpenAI, covering GPT Transcribe: word error rate, speed and price compared with other models.
Grok Voice Transcribe 2.0
Artificial Analysis speech-to-text page for SpaceXAI, covering Grok Voice Transcribe 2.0: word error rate, speed and price compared with other models.
granite-4.0-1b-speech
by IBM Granite
Hugging Face model card for ibm-granite/granite-4.0-1b-speech, IBM's compact speech recognition and translation model.
IndicVoices
by AI4Bhārat
IndicVoices GitHub repo (AI4Bharat): 12,000-hour dataset of natural speech in Indian languages.
Kokoro TTS
by Hexgrad
Hugging Face model card for hexgrad/Kokoro-82M, an 82M-parameter open text-to-speech model, with weights, usage and license details.
Kyutai (Moshi)
by Kyutai
Kyutai: French open-science AI lab releasing audio models such as Moshi, Hibiki and Unmute, with papers.
MAI-Transcribe-1.5
Artificial Analysis speech-to-text page for Microsoft, covering MAI-Transcribe-1.5: word error rate, speed and price compared with other models.
MedASR (Google)
Google Research post on MedGemma 1.5 medical image interpretation and MedASR medical speech-to-text.
Mozilla Common Voice
by Common Voice
Mozilla Common Voice GitHub repo: platform code for the crowdsourced multilingual voice dataset.
NVIDIA Canary-Qwen-2.5B
by NVIDIA
Hugging Face model card for nvidia/canary-qwen-2.5b, NVIDIA's 2.5B speech recognition model combining Canary and Qwen, with weights, usage and license details.
NVIDIA Parakeet TDT 0.6B v3
by NVIDIA
Hugging Face model card for nvidia/parakeet-tdt-0.6b-v3, NVIDIA's 0.6B Parakeet multilingual speech recognition model, with weights, usage and license details.
OpenMOSS (Fudan)
OpenMOSS GitHub organisation (Shanghai Innovation Institute and Fudan): MOSS, MOSS-TTSD and SpeechGPT code.
DeepFilterNet
DeepFilterNet GitHub repo: deep filtering noise suppression for speech audio.
pyannote Hugging Face organization
by pyannote
pyannote Hugging Face organisation: speaker diarization and voice activity detection models.
Resonant-1
Artificial Analysis speech-to-text page for Reson8, covering Resonant-1: word error rate, speed and price compared with other models.
Scribe v2
Artificial Analysis speech-to-text page for ElevenLabs, covering Scribe v2: word error rate, speed and price compared with other models.
Silero models
Silero models GitHub repo: pretrained speech-to-text and text-to-speech models.
Silero VAD
Silero VAD GitHub repo: pretrained voice activity detection model for streaming audio.
smallest-python-sdk
by Smallest AI
Smallest AI Python SDK GitHub repo: official client for the Smallest AI speech API.
Smallest AI Pulse
Artificial Analysis speech-to-text page for Smallest.ai, covering Smallest AI Pulse: word error rate, speed and price compared with other models.
Smallest AI Pulse Pro
Artificial Analysis speech-to-text page for Smallest.ai, covering Smallest AI Pulse Pro: word error rate, speed and price compared with other models.
Solaria-3
Artificial Analysis speech-to-text page for Gladia, covering Solaria-3: word error rate, speed and price compared with other models.