A focused feed of models, agents, research and open-source releases for people building with AI.
HoppouAI released Breeze-TTS-2.cpp , GGUF quants plus a C++ runtime for BreezeBlue's 3B TTS model. Runs on NVIDIA, AMD, Intel GPUs via Vulkan, with CPU fallback and no Python required. Hits roughly 1.2x realtime at Q8_0 on an RTX 3060, 24 kHz bilingual English and Mandarin output.
Suno Studio gets a redesigned, larger Clip Settings sidebar with volume, pitch, loop, reverse, and timestretch. Lyrics, style tags, and stem separation moved to a separate Clip Info panel to reduce clutter.
Hume launched the Audio Decision API , one model that answers arbitrary questions about audio with probabilities. Replaces 10+ specialized audio models (diarization, emotion, laughter, language ID) with a single call.
Edge0 released Audio8-TTS-Realistic-Air , a 0.6B TTS model with zero-shot voice cloning, Apache 2.0 licensed. Supports 23 languages including Arabic, Hindi, Thai, Vietnamese, adding 12 to the previous Preview release. Scores 1.25% error on Seed-TTS, competitive with 4B to 8.5B parameter rivals.
Pika Labs released Pika H3 Flex , open LoRA adapters that run MiniMax H3 in 6 sampling steps instead of 50. Two adapters: FL2VA for text and first/last frame, Ref2VA for up to 12 reference files (images, video, audio).
Haas-ja-v1.0 hits 6.12% CER strict on Japanese ASR, best among open models evaluated. Costs roughly $0.003 per audio hour, about 1/70th of GPT-Transcribe at similar accuracy. Based on NVIDIA Parakeet-TDT with Kimi Delta Attention replacing the LSTM prediction network.
Liquid AI released LFM3 , six open-weight 2B-class models sharing one backbone across text, vision, audio, omni, and decision tasks. New tri-hybrid architecture mixes 13 gated convolutions, 5 sliding-window attention, and 6 global attention layers across 24 layers.
NVIDIA released NemotronLabs RE-USE , a 9.6M parameter open multilingual speech enhancement model. Beats 250M and 500M parameter rivals on PESQ, STOI, speaker similarity, and ASR accuracy. Handles noise, reverb, clipping, codec artifacts, packet loss, and bandwidth limits across 7 sample rates.
Google Gemma team debuted Reachy on Wheels at PyTorch Conference , battery-powered with no cloud. Gemma 4 E4B handles voice conversation and tool calls to start or stop the vehicle. EmbeddingGemma 2 runs on an Arduino Ventuno Q board for real-time camera-based steering.
Community-built Windows desktop app runs Boson AI's Higgs Audio v3 TTS locally via GGUF and native C++/CUDA. Five quant tiers from q4_k_m (8 GB VRAM) to bf16 (16 GB VRAM), with q8_0 as the recommended default.
Maya Research released Maya 2 Native , open-weight streaming TTS for 11 Indic languages plus English under Apache-2.0. 2.5B parameters, 20 built-in voices, zero-shot voice cloning from 3 to 10 seconds of reference audio.
Liquid AI released LFM3 , six open-weight ~2B models sharing one tri-hybrid backbone. Lineup covers text, vision+video, audio, omni, and typed-output decision models ( d1-Omni-2B ). Tri-hybrid: 13 gated short convs, 5 sliding-window attention, 6 global attention layers across 24 layers.
Phonic 1 (High) debuts at #1 on the Speech to Speech Index at 87.1, edging out Grok Voice 3.0. Leads τ-Voice agentic benchmark at 79.9%, a 5.4 point gap over GPT-Live-1 (Astra, medium). Scores 98.9% on Big Bench Audio reasoning, third behind StepAudio 3 and Qwen Audio 3.0.
Fish Audio released Drama 3 , a 35B MoE TTS model with 3B active parameters, open weights on Hugging Face. Ranks #1 open-weights on both Artificial Analysis Speech Arenas and #1 overall on Pronunciation Robustness at 94.6%.
AtomicChat released GGUF builds of NVIDIA's Nemotron 3.5 Transcribe 30B-A4B speech model. Seven decoder sizes from 13.57 GB to 35 GB cover 16 GB GPUs through workstations. Accuracy stays within ±0.18 WER points of BF16 on English, German, French, Spanish.
Suno's new Model Context Protocol server lets Claude, ChatGPT, Cursor, Gemini, and Claude Code generate songs inline without API keys or tab-switching.
A community port by the handy-computer team converts NVIDIA's multilingual Parakeet-TDT-0.6B-v3 into GGUF files for transcribe.cpp, enabling local CPU and GPU inference.
AssemblyAI's first text-to-speech model streams audio in 47 ms, prices at $15 per million characters, and ships with 4,000 voices plus instant cloning.
NVIDIA's open-weight 30B mixture-of-experts speech model handles transcription, diarization, translation and audio Q&A in 40 languages from one checkpoint.
NVIDIA's new 30B mixture-of-experts transcription model matches top open-weight accuracy at 2.8% WER while using 6x fewer active parameters than rivals.
SYSTRAN's English-only base checkpoint repackages OpenAI Whisper for CTranslate2, delivering roughly 4x faster transcription at the same accuracy on modest hardware.
Fish Audio's new streaming speech-to-text claims the top spot on Cekura's leaderboard with 2.21% WER and conversation-aware turn-taking for voice agents.
Tencent open-sourced a 30B MoE model for live Chinese-English speech interpretation, hitting roughly 3-second latency with real-time voice cloning.
InclusionAI drops five Ming-3.0 speech models under MIT, led by a 7.9B MoE TTS that tops open-weight controlled voice leaderboards.
No stories match these filters.
