Speech Recognition and TTS in less than 500kb

Moonshine AI released Moonshine Micro, an open-source AI toolkit for real-time voice applications on embedded systems, running in as little as 470 KB of RAM on the 80-cent Raspberry Pi RP2350 microcontroller. The toolkit includes voice activity detection, command recognition, and neural speech synthesis, enabling voice interfaces on resource-constrained devices.

Moonshine Voice https://github.com/moonshine-ai/moonshine is an open source AI toolkit for developers building real-time voice agents and applications. Moonshine Micro is a version designed specifically for embedded system processors like microcontrollers and DSPs, and uses the Raspberry Pi RP2350, which retails for just 80 cents, as its reference platform. It includes voice-activity detection /moonshine-ai/moonshine/blob/main/micro/vad/README.md , command recognition /moonshine-ai/moonshine/blob/main/micro/stt/README.md , and neural speech synthesis /moonshine-ai/moonshine/blob/main/micro/neural-tts/README.md and can run in as little as 470 KB of RAM. You can see a full walkthrough in the video below: The memory and compute requirements are designed to fit resource-constrained systems. Figures below are for the RP2350 demo /moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md ; the detailed memory budget /moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md memory-budget breaks each one down: | Component | Flash | SRAM arena peak | Compute | |---|---|---|---| | VAD Voice Activity Detection | ~89 KiB | ~36 KiB | ~0.8 MMAC/frame ~25 MMAC/s | | STT SpellingCNN Speech-to-Text | ~1.3 MiB | ~346 KiB | ~36 MMAC/s | | TTS neural diphone synth @ 16 kHz | ~1.8 MiB voice pack | ~340 KiB | ~37 MMAC typical reply ~65 MMAC/s out | TOTAL Demo pipeline | ~3.6 MiB | ~468 KiB provisioned | classify + speak ~0.7–1.0 s | Notes: Flash is .text + .rodata measured with arm-none-eabi-size on the default moonshine micro echo firmware includes the embedded neural voice pack ; SRAM is .bss + heap + stacks. VAD, STT, and neural TTS run sequentially and time-share one ~384 KiB TFLM arena, so SRAM is not additive — ~468 KiB is the total RAM provisioned on the 520 KiB RP2350 wifi hardware ~491 KiB . A MAC is one multiply-accumulate; MMAC/s = millions per second during the active non-idle stage. The code is released under the permissive MIT License license , suitable for commercial applications. There's a complete end-to-end example /moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md showing how to set up a wifi connection on a microcontroller using voice on an RP2350 MCU. The VAD, STT, and TTS libraries can be used independently of each other, relying on the included TensorFlow Lite Micro https://github.com/tensorflow/tflite-micro library for the neural computations. Voice Activity Detection /moonshine-ai/moonshine/blob/main/micro/vad/README.md Speech to Text /moonshine-ai/moonshine/blob/main/micro/stt/README.md Custom Word Recognition /moonshine-ai/moonshine/blob/main/micro/stt-training/README.md Neural Text to Speech /moonshine-ai/moonshine/blob/main/micro/neural-tts/README.md Wifi Setup Example /moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md This code, apart from the source in third-party/ , is licensed under the MIT License — see LICENSE /moonshine-ai/moonshine/blob/main/micro/LICENSE in this directory also at the repository root . The SpellingCNN and TinyVadCNN models in models/ /moonshine-ai/moonshine/blob/main/micro/models are released under the MIT License. The code in third-party/ is licensed according to the terms of the open source projects it originates from, with details in a LICENSE file in each subfolder.