Voice on mobile with interruption using Apple’s SpeechAnalyzer, MLX Swift for running the LLM and the following CoreML models: Kokoro, Silero and SegmentAnyText. All running in orchestration. Awni Hannun Fluid Inference

The future of AI may fit in your hand. Fluid Inference is bringing AI from the cloud to your devices with an open-source platform for local transformer optimization. CEO Brandon Weng (brandon) joins #IntelOnAI to talk cost, speed, and equity. Listen now. Apple:

👀👀

#iOSdev tool alert! 🚨 FluidAudio by is a high-performance Swift framework for on-device speaker diarization and audio processing, designed to meet research standards. github.com/FluidInference…

we have officially crossed 900 stars on fluidaudio, with this current rate of growth, we can expect 1000 before 2026! github.com/FluidInference…

lfg

Custom vocabulary support is now available in FluidAudio from Fluid Inference

streaming transcription with parakeet is almost instant s/o Alex Fluid Inference

appreciate your help and input here 🙏 love the community we have

👏

Lately it’s been a lot easier converting models such as pocket-tts or nemotron to coreml thanks to mobius. I think after it had gained sufficient examples such as nvidia parakeet or streaming models conversion scripts. Coding agents are able to take a look at the process and

Alex And here it is, kyutai PocketTTS on mobile.

i just finally got qwen-asr converted to cormel. what was surprisingly was that it uses a encoder -> embedder -> llm transcription. the architecture is quite alien from traditional nvidia transformer transducer pipelines. this is something i am unfamiliar with but it does mean

we now support kyutai-lab's pocket tts in FluidAudio. PocketTTS is a streaming TTS that handles long text with seamless chunk transitions and is a more permissive model for commercial usages . Get started here: github.com/FluidInference… credits to Sachin Desai for helping us test

beautiful

🫡