Add ASIO monitor split proof

This commit is contained in:
Meta Vens 2026-05-28 08:21:08 +01:00
commit c192eba35a
5 changed files with 83 additions and 1 deletions

View file

@ -0,0 +1,53 @@
# Complex Contour Monitor Split Receipt 2026-05-28 08:18:13
## Setup
- Renderer: `canary-packet-trill`
- Playback artifact: `artifacts/asio/canary-packet-monitor-split-20260528-081813-f32.raw`
- Output 0 capture: `artifacts/asio/scarlett-canary-packet-monitor-output0-20260528-081813-f32.raw`
- Output 1 capture: `artifacts/asio/scarlett-canary-packet-monitor-output1-20260528-081813-f32.raw`
- ASIO driver: `Focusrite USB ASIO`
- Sample rate: `192000`
- Capture seconds: `4.2`
- Probe change: `--play-output-channel` selects one ASIO output buffer; `-1`
remains the default all-output behavior.
- Physical mapping:
- `asio-ch0` / Input 1: shotgun, about 1.5 m from right monitor and 3.0 m from left monitor
- `asio-ch1` / Input 2: cardioid, about 0.5 m from left monitor and 1.5 m from right monitor
## Signal Sanity
Output 0 energized loopback channel `asio-ch2`; `asio-ch3` was silent.
Output 1 energized loopback channel `asio-ch3`; `asio-ch2` was silent.
Physical input RMS stayed high in both captures:
| Capture | `asio-ch0` RMS | `asio-ch1` RMS | Reference loopback |
| --- | ---: | ---: | --- |
| output 0 | `0.014660` | `0.050159` | `asio-ch2` |
| output 1 | `0.019050` | `0.040859` | `asio-ch3` |
## Results
| Output | Reference | Mic | Estimate | Confidence | Direct Hits |
| --- | --- | --- | ---: | ---: | ---: |
| output 0 | `asio-ch2` | shotgun / `asio-ch0` | `783.629083` samples / `4081.401 us` | `0.392` | `19` |
| output 1 | `asio-ch3` | shotgun / `asio-ch0` | `779.619141` samples / `4060.516 us` | `0.376` | `18` |
| output 0 | `asio-ch2` | cardioid / `asio-ch1` | `547.779779` samples / `2853.020 us` | `0.285` | `14` |
| output 1 | `asio-ch3` | cardioid / `asio-ch1` | `543.020433` samples / `2828.231 us` | `0.383` | `17` |
## Interpretation
The ASIO probe now isolates output buffers, and the loopback channels confirm
that output 0 and output 1 are distinct at the interface. The physical mic
delays, however, are nearly identical across output channels:
- shotgun differs by about `20.885 us`;
- cardioid differs by about `24.789 us`.
That is not compatible with the declared monitor geometry if output 0 and
output 1 correspond to isolated left/right monitors in the room. The current
evidence says the downstream monitor path is probably summing, mirroring, or
otherwise not isolating the physical speakers as assumed. Do not use this
receipt to learn left/right acoustic geometry until physical speaker isolation
is verified.

View file

@ -125,6 +125,7 @@ struct CaptureState
std::vector<unsigned long long> sweepSamples;
std::vector<float>* playbackSamples = nullptr;
double playbackGain = 1.0;
long playbackOutputChannel = -1;
std::vector<float>* recordedInput = nullptr;
unsigned long long recordFrames = 0;
unsigned long long sweepFrame = 0;
@ -320,6 +321,11 @@ void processOutput(long doubleBufferIndex)
for (size_t channel = 0; channel < g_capture.outputBuffers->size(); ++channel)
{
if (g_capture.playbackOutputChannel >= 0 && static_cast<long>(channel) != g_capture.playbackOutputChannel)
{
continue;
}
auto& buffer = (*g_capture.outputBuffers)[channel];
auto& info = (*g_capture.outputChannels)[channel];
auto* raw = static_cast<unsigned char*>(buffer.buffers[doubleBufferIndex]);
@ -498,6 +504,12 @@ double doubleOption(int argc, char** argv, const char* name, double fallback)
return raw ? std::atof(raw) : fallback;
}
long longOption(int argc, char** argv, const char* name, long fallback)
{
const char* raw = option(argc, argv, name, nullptr);
return raw ? std::atol(raw) : fallback;
}
bool hasArg(int argc, char** argv, const char* name)
{
for (int index = 1; index < argc; ++index)
@ -576,6 +588,7 @@ int main(int argc, char** argv)
const char* playFloat32MonoPath = option(argc, argv, "--play-f32-mono", nullptr);
const char* recordFloat32InterleavedPath = option(argc, argv, "--record-f32-interleaved", nullptr);
const auto playGain = doubleOption(argc, argv, "--play-gain", 1.0);
const auto playOutputChannel = longOption(argc, argv, "--play-output-channel", -1);
const auto sweepGain = doubleOption(argc, argv, "--sweep-gain", 0.03);
const auto sweepToneSeconds = doubleOption(argc, argv, "--sweep-tone-seconds", 0.55);
const auto sweepGapSeconds = doubleOption(argc, argv, "--sweep-gap-seconds", 0.20);
@ -731,7 +744,7 @@ int main(int argc, char** argv)
playbackSamples.resize(static_cast<size_t>(byteCount) / sizeof(float));
input.read(reinterpret_cast<char*>(playbackSamples.data()), static_cast<std::streamsize>(playbackSamples.size() * sizeof(float)));
}
std::printf("asio playback-f32-mono: path=%s samples=%zu gain=%.5f\n", playFloat32MonoPath, playbackSamples.size(), playGain);
std::printf("asio playback-f32-mono: path=%s samples=%zu gain=%.5f outputChannel=%ld\n", playFloat32MonoPath, playbackSamples.size(), playGain, playOutputChannel);
}
if (captureSeconds > 0.0 || monitorSweep || !playbackSamples.empty() || emitJsonBlocks)
@ -820,6 +833,7 @@ int main(int argc, char** argv)
g_capture.sweepGain = sweepGain;
g_capture.playbackSamples = playbackSamples.empty() ? nullptr : &playbackSamples;
g_capture.playbackGain = playGain;
g_capture.playbackOutputChannel = playOutputChannel;
g_capture.emitJsonBlocks = emitJsonBlocks;
g_capture.sweepFrame = 0;
g_capture.sweepEnergy.assign(sweepFrequencies.size() * static_cast<size_t>(inputs), 0.0);

View file

@ -314,6 +314,13 @@ the shotgun, about 1.5 m from the right monitor and 3.0 m from the left;
`asio-ch1` is the cardioid, about 0.5 m from the left monitor and 1.5 m from
the right. Repeat captures are needed before updating the channel model or
letting physical mic reports drive actuator authority.
`complex-contour-monitor-split-20260528-081813.md` adds
`--play-output-channel` to the ASIO probe and captures output 0 and output 1
separately. Loopback proves the ASIO output buffers are distinct, but physical
mic delays are nearly identical across outputs: shotgun differs by about
20.885 us and cardioid by about 24.789 us. That does not match the declared
left/right monitor geometry, so the downstream monitor path is probably
summing, mirroring, or otherwise not isolating the physical speakers.
The first actuator proof now exists: `faust/mimir_alignment_actuator.dsp` owns
six channels of bounded fractional delay/gain controls for Faust/native DSP,
and `Mimir.BufferSmoke --bioacoustic-actuator-self-test --sample-rate 48000

View file

@ -203,3 +203,4 @@
{"ts": "2026-05-28T00:27:19+01:00", "type": "verification", "status": "blocked", "note": "Fresh complex-contour Scarlett proof 20260528-002313: ASIO loopback ch2->ch3 measured -0.062 us with 0.956 confidence and 257 direct hits, but physical inputs ch0/ch1 produced zero contour hits against 1480 reference hits. Physical mic proof is blocked on input signal/gain/routing/source placement, not receiver code."}
{"ts": "2026-05-28T08:09:07+01:00", "type": "verification", "status": "ok", "note": "Monitor-routed complex-contour Scarlett proof 20260528-080727: loopback ch2->ch3 measured -0.014 us at 0.971 confidence; physical Input 1/asio-ch0 landed -0.410 us from seed at 0.524 confidence with 24 direct hits; Input 2/asio-ch1 landed -2.467 us at 0.363 confidence with 14 direct hits. Previous 080604 run was headphone-routed by mistake and is marked as misroute witness."}
{"ts": "2026-05-28T08:14:31+01:00", "type": "observation", "status": "ok", "note": "Physical mapping for monitor-routed complex-contour proof: asio-ch0/Input 1 is the shotgun, about 1.5 m from right monitor and 3.0 m from left; asio-ch1/Input 2 is the cardioid, about 0.5 m from left monitor and 1.5 m from right. Geometry should be used when interpreting 20260528-080727 path delays."}
{"ts": "2026-05-28T08:20:48+01:00", "type": "verification", "status": "blocked", "note": "ASIO monitor split proof 20260528-081813 added --play-output-channel and captured output 0/output 1 separately. Loopback channels confirmed distinct ASIO output buffers, but physical mic delays were nearly identical across outputs (shotgun delta about 20.885 us; cardioid delta about 24.789 us), inconsistent with declared left/right monitor geometry. Downstream monitor routing/isolation must be verified before learning left/right acoustic paths."}

View file

@ -328,6 +328,13 @@ accepted_design:
from the left; `asio-ch1` is the cardioid, about 0.5 m from the left
monitor and 1.5 m from the right. Repeat captures are needed before
channel-model updates or actuator authority.
`complex-contour-monitor-split-20260528-081813.md` added ASIO
`--play-output-channel` and captured output 0/output 1 separately.
Loopback channels prove the ASIO output buffers are distinct, but
physical mic delays differ by only about 20.885 us for the shotgun and
24.789 us for the cardioid across outputs. This does not match the
declared monitor geometry; verify downstream monitor routing/isolation
before learning left/right acoustic paths.
`calibration/bioacoustic/complex-contour-replay-panel.json` persists the
current four-case replay receipt.
Acoustic robustness remains open.