FireRedAudio
One open 9B audio model for speech recognition, understanding, TTS and editing
Emerged finds new products worth a look, daily — what’s new today
Nari Qwen3-TTS and Qwen3-ASR
Fast, low-cost speech APIs built on Qwen3 voice models
Voice agent builders get top-ranked accuracy and latency on Coval's benchmarks at a fraction of ElevenLabs or Cartesia pricing.
Nari Labs runs its own inference engines for Qwen3's 1.7B text-to-speech and speech-recognition models and sells them as APIs. On Coval's benchmarks the TTS endpoint ranks first for word error rate and the ASR endpoint first for latency, at 44ms median. TTS starts at $5 per million characters and ASR at $0.06 an hour. Both are in public beta with free trial credits, and the TTS inference engine is open source.
More like this
Lokutor
Voice agents that run entirely on CPUs, in the cloud, on-prem or on-device
Nico
Hold a key, say what you need, and a sandboxed worker on your Mac does it
VoiceMem
Open streaming memory for real-time voice agents, split into facts and feelings
Sparrow-2
A turn-taking model that knows when a voice agent should speak
Aloud
A Mac menu-bar reader that speaks any text with the current word highlighted