AI tools for Speech To Speech

Phonic-1 is a hosted speech-to-speech model for voice agents. It receives the caller's audio directly and responds in spoken audio, using cues such as pace and hesitation to sup

MOSS-Live is a real-time conversational system from OpenMOSS. It accepts live speech and visual context and can keep listening while it speaks, making interruptions and topic ch

Audio8-TTS-Realistic-Air is a compact speech-generation model with multilingual synthesis and zero-shot voice cloning. Its 0.6-billion-parameter main model supports 23 languages

Hy-SimulMT is Tencent Hunyuan's real-time simultaneous translation model for Chinese and English. It processes speech as it arrives and produces incremental translated text, wit

NVIDIA Nemotron 3.5 Transcribe is an open-weight speech and audio understanding model. It combines recordings, instructions, and optional context to produce text for multilingua

Dessa is Block's full-duplex conversation model, trained directly on human speech to hold expressive real-time conversations. It learns from audio without a text-first transcrip

Ming-3.0-TTS is an open-weight speech-generation family for narration, audio stories, podcasts, and conversational products. A lightweight dense model and a mixture-of-experts v

Drama 3 is Fish Audio's expressive voice-generation model for directing how a line is performed, not just which words are spoken. Natural-language instructions and inline contro

Underdog Bark 3.0 is a compact English speech recognition family for private dictation and meeting transcription. The model runs locally on a Mac, with small downloadable varian

Eleven v4 is ElevenLabs' expressive speech-model family, including a Turbo variant for real-time voice applications. The model interprets script context and inline audio tags to

Whistle is Cactus's compact speech-recognition model for phones, wearables, and other CPU-based devices. A 16.9 MB quantized model runs in the same C++ engine as Needle and can

Phonon-2 is a compact open English speech-recognition model for local transcription. Its encoder stores weights in five learned levels at approximately 2.1 bits, compressing a P

Gemini 3.8 TTS is Google's expressive speech-generation family, including Flash TTS and Flash-Lite TTS. Natural-language directions guide character voices and line-by-line deliv

Sophea AI Meet is a meeting intelligence platform that records conversations and turns them into searchable transcripts, summaries, decisions, and follow-up tasks. Its transcrip

AuK is a speech foundation model that generates and edits audio through natural-language instructions. It combines speech synthesis, spoken-content changes, enhancement, separat

VoiceMem adds persistent memory to voice agents by organizing factual information alongside preferences, personality, and emotional context. It can ingest speech and retrieve re

Gemini 3.5 Transcribe turns live or recorded speech into formatted text with support for vocabulary adaptation and conversational cleanup. It is designed for voice interfaces, c

Audio8 TTS Preview 0.1B is a compact zero-shot text-to-speech model for speech generation and voice cloning. Its main generative model is approximately 170 million parameters, w

MiDashengLM-Gen is an end-to-end framework for unified audio scene generation. Given a structured text description, it can generate a coherent scene that blends speech, music, s

IndexTTS 2.5 is a zero-shot neural text-to-speech foundation model for expressive multilingual voice synthesis. It builds on a transformer-based Text-to-Semantic module and a no