whereismyai
Esp32 Pie ASR
Esp32 Pie ASR: Live English speech recognition runs on the ESP32-P4 microcontroller without cloud connectivity.
This project implements live English speech recognition directly on the ESP32-P4 microcontroller using a 65.7M-parameter streaming Zipformer2 transducer model. The system quantizes weights to 2 and 4 bits, storing them on a microSD card and loading them into PSRAM at boot. It performs bit-exact computations matching PC builds using RISC-V cores and PIE SIMD assembly kernels. The software supports open-vocabulary streaming transcription and includes voice-control demos driven by word rules. Supported hardware inputs and outputs include the board microphone and screen.
Esp32 Pie ASR at a glance
Key features
- Live English speech recognition
- Open vocabulary streaming
- Faster than real-time speed
- On-device execution
- Phrase and hotword biasing
- Voice control demos
- Bit-exact PC parity
Use it for
- Live speech transcription on microcontrollers
- Voice control for smart home devices
- Audio command processing for robot vacuums
- Embedded speech recognition testing
Alternatives to Esp32 Pie ASR
- Whistle Esp32 — Transcribes speech and executes commands locally on a ten-dollar microcontroller.
- Free ASR — Zero-dependency real-time streaming speech-to-text with no account login required.
- AssemblyAI — Industry-leading speech models deliver high accuracy and low latency for voice applications.
- sherpa-onnx — Runs complex audio AI tasks offline on local devices and embedded hardware.
- whisper.cpp — It provides high-performance, offline speech recognition on diverse hardware.
- Esp32 Thinks — Runs your own AI stack on low-cost hardware without vendor lock-in.
- Piper — Runs locally and offers fast neural text-to-speech synthesis.
- WhisperLive — It provides real-time speech-to-text using multiple backends like Faster Whisper and TensorRT.
- Babytalk — Runs entirely on-device with no cloud or internet required.
- WeNet — Achieves state-of-the-art results on public speech datasets.
- Pipecat — Builds real-time voice and multimodal agents with frame-level control.
- Esp32 AI — Runs a language model directly on a microcontroller chip without a server.
All Esp32 Pie ASR alternatives