mlcommons / mlcommons/tiny

No wakeword detection on Streaming Wakeword benchmark.

Open
#175 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C
Stars
479
Forks
116
PR merge metrics
No merged PRs in 30d

Description

Hello, I have gotten the new runner to run the streaming wakeword benchmark on energy mode using the reference submission on the L4R5ZI reference board, successfully receiving power and performance measurements. However, I am not getting any actual detections. It seems like the HAL_SAI_RxCpltCallback() is never being triggered. I have verified that my GPIO configuration and physical wiring are correct as well. Do you have any suggestions?

Here is a snippet of my output:
Device name retrieved: ML Commons
Device profile retrieved: MLPerf Tiny Firmware V0.1.0
Device model retrieved: sww01
Playing short_wav_easy10s.wav (length = 10.0s) at 16:32:01...
... done
[]
samples:2382 timestamps:2
SerialDevice exiting port /dev/cu.usbmodem114403, entry_count (before decrement) = 1
pwr pwr off
PM: stop
Done with PM.stop()
pwr pwr off
pwr hrc
SerialDevice exiting port /dev/cu.usbmodem2078306750341, entry_count (before decrement) = 1
Power Edition Output
1 iterations. Elapsed time = 0.30100000000000016,
total energy = 0.0312331
Per inference: time = 0.30100000000000016, energy = 0.0312331

Thanks in advance!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the streaming wakeword benchmark and the HAL_SAI_RxCpltCallback() path on the L4R5ZI reference board, using the reference submission and the reported output as the reproduction case. Confirm why the callback is not triggered despite the GPIO configuration and wiring checks; done means the benchmark produces actual wakeword detections as well as power and performance measurements.

Written by the indexing model from the issue text.

Assessment

Tech stack
c
Domain
embedded-iot, testing-qa
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.