Independent public record
Scott Mills
Audio engineer and software engineer, Saskatchewan, Canada
- Document
- Public consultation submission
- To
- Innovation, Science and Economic Development Canada
- Filed
- 20 September 2026
- Window
- 23 July to 23 September 2026
- Status
- Public document. May be posted by ISED.
The finding
Present audio detection should not be the enforcement basis for a per-work labelling mandate. It performs well on clean, fully-generated tracks and degrades sharply on hybrid and post-processed material — which is precisely the case ISED's own discussion paper puts at the centre of this question.
The rest of the submission depends on that finding. Hybrid and post-processed music is the case the discussion paper names. Detection and watermarking do not hold there. Per-component attribution of a full-mix generator output is a category error: those systems never produce a per-instrument representation.
Who is submitting this
I am an audio engineer and software engineer in Saskatchewan. I have used generative music tools at high volume — thousands of tracks generated and released through commercial distribution — and I spent the past year attempting to verify, attribute, decompose, and transcribe the results, which is what produced the submission.
Until early 2026 I moderated a Suno practitioner community of approximately 50,000 members. I no longer hold that role. Observations drawn from that experience are labelled where they appear. I can describe what that population does with these tools, and I can serve as a research subject. I cannot convene it on the government's behalf, because I am no longer part of it.
Instruments named in the submission
Two of the tools described in the accompanying field report are public and inspectable.
- tab-agent-pro
Named in the submission. Public and inspectable. Audio-to-tablature for guitar and bass.
- stem-midi-pro
Named in the submission. Public and inspectable. Stem separation and MIDI from mixed audio.
Other public software
- sgm-suno-workflow — REAPER scripts used in the volume practice described in the submission.
- 30-agents — Self-hosted multi-agent orchestration. Public source.
- apc-mcp — Build and test tools for audio plugin projects.
What I asked ISED to take from this
- Make the Code's detection commitment condition-specific, and publish accuracy by condition.
- Do not condition per-work labelling on per-component attribution of a full mix.
- Prefer process disclosure to decomposition.
- Require the stem-export disclosure from full-mix generators — one sentence, no new instrument.
- Keep training-data manifests in scope, and do not present them as per-output attribution.
- Add provenance loss to the incident categories, with a channel that accepts de-identified reports.
- Treat practitioners as an evidence source, not only as an audience.
Full argument, recommendations, and sources are in the submission.
Contact
I am available to appear, to provide data, to serve as a research subject, or to be contacted for follow-up.