Nari Labs offers Qwen3-TTS and Qwen3-ASR as text-to-speech and speech-to-text APIs aimed at accuracy, latency and cost. The team reports leading Coval September 2026 voice benchmarks on STT latency and TTS word error rate, sitting on the quality-latency Pareto frontier for both directions. The free public beta ended in September 2026.
Be the first to rate Nari Labs. Your feedback helps other users decide.
Run speech-to-text with a focus on low latency
Produce natural speech through the Qwen3-TTS models
Pick models that sit on the quality-cost frontier
Similar tools in the same category — compare side-by-side
Sayso is dictation that runs on your own computer. You press a key, speak, and the text lands in the...
Blokkily is an experimental pattern-first digital audio workstation. The same pattern can be edited ...
Free, open source voice dictation for macOS. Yap does on-device transcription with Apple's Speech fr...
Guides and comparisons about audio tools
Seven narrowly aimed AI launches: tiny model benchmarks in a browser, an open-source trading workspace, analytics that can refuse a query, a Word-editing model.
Murf AI produces clean narration at a friendly price. It stops being enough when you need top-tier naturalness, or when voice was only one step.
ElevenLabs has the most natural speech available and charges by the character. Five alternatives for when that cost, or voice-only scope, is wrong.
Descript's transcript editing is brilliant if your audio has speech in it. Five alternatives for when it doesn't, or when you need visuals instead.
How to create AI-generated podcast show notes in 2026: transcription with Whisper or Descript, then structured notes from ChatGPT or Claude.
