{"slug": "speech-recognition-and-tts-in-less-than-500kb", "title": "Speech Recognition and TTS in less than 500kb", "summary": "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.", "body_md": "[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.\n\nYou can see a full walkthrough in the video below:\n\nThe memory and compute requirements are designed to fit resource-constrained\nsystems. Figures below are for [the RP2350 demo](/moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md); the\ndetailed [memory budget](/moonshine-ai/moonshine/blob/main/micro/examples/rp2350/README.md#memory-budget) breaks each one down:\n\n| Component | Flash | SRAM (arena peak) | Compute |\n|---|---|---|---|\n| VAD (Voice Activity Detection) | ~89 KiB | ~36 KiB | ~0.8 MMAC/frame (~25 MMAC/s) |\n| STT (SpellingCNN Speech-to-Text) | ~1.3 MiB | ~346 KiB | ~36 MMAC/s |\n| TTS (neural diphone synth @ 16 kHz) | ~1.8 MiB voice pack | ~340 KiB | ~37 MMAC typical reply (~65 MMAC/s out) |\nTOTAL (Demo pipeline) |\n~3.6 MiB |\n~468 KiB provisioned* |\nclassify + speak ~0.7–1.0 s |\n\n*Notes:*\n\n*Flash is*`.text`\n\n+`.rodata`\n\nmeasured with`arm-none-eabi-size`\n\non the default`moonshine_micro_echo`\n\nfirmware (includes the embedded neural voice pack); SRAM is`.bss`\n\n+ 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`\n\n~491 KiB).*A MAC is one multiply-accumulate; MMAC/s = millions per second during the active (non-idle) stage.*\n\nThe code is released under [the permissive MIT License](#license), suitable for commercial applications.\n\nThere'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.\n\nThe 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.\n\n[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)\n\nThis code, apart from the source in `third-party/`\n\n, is licensed under the MIT\nLicense — see [LICENSE](/moonshine-ai/moonshine/blob/main/micro/LICENSE) in this directory (also at the repository root).\n\nThe SpellingCNN and TinyVadCNN models in [ models/](/moonshine-ai/moonshine/blob/main/micro/models) are released under\nthe MIT License.\n\nThe code in `third-party/`\n\nis licensed according to the terms of the open\nsource projects it originates from, with details in a LICENSE file in each\nsubfolder.", "url": "https://wpnews.pro/news/speech-recognition-and-tts-in-less-than-500kb", "canonical_source": "https://github.com/moonshine-ai/moonshine/tree/main/micro", "published_at": "2026-07-14 19:25:10+00:00", "updated_at": "2026-07-14 19:48:18.075444+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning"], "entities": ["Moonshine AI", "Moonshine Micro", "Raspberry Pi RP2350", "TensorFlow Lite Micro"], "alternates": {"html": "https://wpnews.pro/news/speech-recognition-and-tts-in-less-than-500kb", "markdown": "https://wpnews.pro/news/speech-recognition-and-tts-in-less-than-500kb.md", "text": "https://wpnews.pro/news/speech-recognition-and-tts-in-less-than-500kb.txt", "jsonld": "https://wpnews.pro/news/speech-recognition-and-tts-in-less-than-500kb.jsonld"}}