The authors built Oído, an inference engine that executes a 13M-parameter Conformer-CTC model on an ESP32-S3 (240 MHz dual-core Xtensa LX7, 512 KB SRAM, 8 MB PSRAM, 16 MB flash).
Open-vocabulary speech recognition runs on a $5 microcontroller without a neural accelerator
Oído, an int8 inference engine on the ESP32-S3, transcribes English from a 13M-parameter Conformer-CTC model at 3.70% word error rate on LibriSpeech test-clean, within 0.12 points of full precision, but at roughly twice real time.
Industry
Daniel Varela Sánchez · Guillermo Martín-Coello Juárez
Lokutor
Research Digest··3 min read
The authors present Oído, an inference engine that runs open-vocabulary English speech recognition entirely on an ESP32-S3 microcontroller, a $5 chip with no neural accelerator.
Why this paper
From Lokutor
In one line
Oído brings open-vocabulary English speech recognition to a $5 microcontroller without a neural accelerator.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ✓Compute or model size stated (params 13M)
- ✓Limitations stated by the authors (3 noted)
- ✓Reports numbers on named benchmarks (5 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§