Rebuilding my local voice assistant pipeline

(blog.platypush.tech)

81 points | by todsacerdoti 8 hours ago ago

14 comments

  • bill10 2 minutes ago ago

    I am using the old whisper.cpp to turn voice into text, and Kokoro to read text out. Works pretty well for me, but Kokoro still sounds AI and whisper is not that accurate. Will try Parakeet and Silero

  • mmmooo 5 hours ago ago

    older amazon echo dots and shows really make great hardware for this, combined with homeassistant, and can be had for very little cost. e.g. shameless plug: https://github.com/ygelfand/echolocal

    • nl 4 hours ago ago

      I just commented in another thread that I built this: https://github.com/nlothian/earwing

      Beam forming microphone array, custom UI, custom case. Maybe costs more than a second hand echo dot though (hardware is probably around $30).

    • neumann 4 hours ago ago

      I got an echo show rooted and ROMd that I am loving - put it in the kids room as a music player in kiosk mode, but have been now wanting one for myself with wakewords. This looks like the ticket.

      • mmmooo 3 hours ago ago

        the gen 1 or 2 show 8s make great devices for this kind of stuff, 5's are a little small for me. the lenovo android things (google homes) are also really nice if you want a slightly bigger screen, but not as cheap second hand.

  • mewse-hn 5 hours ago ago

    Interesting write up, I've been vibe coding an android voice assistant using a similar pipeline without a wakeword - VAD > ASR > LLM > TTS

    I've found parakeet to be much, much better than whisper and I made a 4 bit quant of parakeet 110M for my crappier android test devices, it would probably fit a raspberry pi:

    https://huggingface.co/mewse/parakeet-tdt_ctc-110m-onnx-int4

    • deepsquirrelnet 26 minutes ago ago

      I trained a streaming model that comes with a VAD. Complete with benchmarks on a pi 4.

      https://huggingface.co/spaces/futo-org/asr4all-client-side

      Large is feasible on one core of a pi 4, but not in low latency streaming. All will work well on a 5.

      I have built exactly the pipeline you're talking about with this model and gemma e4b + mtp. Works entirely on CPU at very reasonable speeds. With an LLM you can even use lighter models and it has negligible impact.

    • GaggiX 4 hours ago ago

      There is also parakeet-redux, a 1.58bits quant of parakeet-tdt-0.6b-v3 that runs as well as the original model: https://huggingface.co/moondream/parakeet-redux

      The model is actually a distillation of a different checkpoint called parakeet-ultra, same size and architecture as parakeet-tdt-0.6b-v3 but better, that's why the 1.58bit quant works so well.

      • mewse-hn 4 hours ago ago

        Oh crazy I hadn't seen this yet, I might be able to unify my 0.6B vs 110M selector, thanks!

  • fragilerock 7 hours ago ago

    Interesting read! I have my own local voice assistant and I implemented exactly the same question mark trick just the other day.

    The overall pipeline is quite similar to my previous setup, but I found the STT -> LLM -> TTS pipeline just infuriatingly slow. TTS and STT with whisper/piper worked fine and reasonably fast for me locally, but the real bottleneck was always the fact that they're not running concurrently with the LLM call. I ended up forking an open source openai realtime API ESP32 plugin and patching it to my liking. It's a bit more expensive in terms of API cost but I only use it sparingly and it's so much nicer to use now.

  • bityard 5 hours ago ago

    Is this complimentary to, or an alternative to Home Assistant?

    • malfist 5 hours ago ago

      Home assistant doesn't really provide much out of the box, its mostly a modular/pluggable ecosystem and one of the things you can install are various voice assistants

  • dofm 5 hours ago ago

    Just what I want to read — favourited to come back to!

  • stavros 7 hours ago ago

    This is really useful, thanks! The other day I bought a small screen and didn't know what to do with it so I made an assistant that talks to my bot:

    https://m.youtube.com/shorts/3FseVyAG5uM

    I used microwakeword too, and training my own wakeword was pretty simple (basically just recording a few tens of samples). Highly recommended.

    I also wanted to be able to issue some voice commands for lights and other things in my home with low latency, so I added a short-circuit plugin that runs right after recognition. If the recognized command matches one of the regexes, it's immediately executed, which has a latency of under a second. Otherwise, it gets sent to my assistant, which usually will tweak the regex (or add a new one) to improve the whole system as it fails.