Manuel/meeting-assistant
Public Access
1
0
Fork 0
Code Issues Pull Requests 4 Actions Packages Projects Releases Wiki Activity Open in Coder

Add inactivity safeguard and speaker diagnostics
PR and Push Build/Test / build-and-test (push) Failing after 8m31s
Details

Browse Source
This commit is contained in:
codex
2026-06-02 13:16:02 +02:00
parent c56ecb6ab3
commit 0e9d525b63
24 changed files with 1780 additions and 117 deletions
Download Patch File Download Diff File Expand all files Collapse all files
MeetingAssistant/Speakers/PyannoteSpeakerIdentityMatchValidator.cs
+70 -15
View File
@@ -30,14 +30,29 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
}
var segments = await DiarizeBytesAsync(wavBytes, cancellationToken);
var result = HasSingleSpeaker(segments);
if (!result)
var result = AnalyzeSingleSpeakerCoverage(segments);
if (!result.Accepted)
{
logger.LogInformation(
"pyannote rejected speaker identity sample because it did not contain one dominant speaker");
"pyannote rejected speaker identity sample because it did not contain one dominant speaker: segments {SegmentCount}, speakers {SpeakerCount}, dominant speaker {DominantSpeaker}, coverage {Coverage:P2}, required coverage {RequiredCoverage:P2}, duration {Duration}",
segments.Count,
result.SpeakerCount,
result.DominantSpeaker,
result.Coverage,
Math.Clamp(options.MinimumSingleSpeakerCoverage, 0, 1),
result.TotalDuration);
return false;
}
return result;
logger.LogInformation(
"pyannote accepted speaker identity sample: segments {SegmentCount}, speakers {SpeakerCount}, dominant speaker {DominantSpeaker}, coverage {Coverage:P2}, duration {Duration}, sample bytes {SampleBytes}",
segments.Count,
result.SpeakerCount,
result.DominantSpeaker,
result.Coverage,
result.TotalDuration,
wavBytes.Length);
return true;
}
public async Task<bool> ValidateMatchAsync(
@@ -63,10 +78,20 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
TimeSpan.FromSeconds(1));
if (layout.UnknownSegment is null || layout.KnownSegments.Count == 0)
{
logger.LogInformation(
"pyannote rejected speaker identity match for {DiarizedSpeaker} and identity {IdentityId} because no readable composite snippets were available",
request.DiarizedSpeaker,
request.IdentityId);
return false;
}
var segments = await DiarizePathAsync(tempPath, cancellationToken);
logger.LogInformation(
"pyannote diarized speaker identity match for {DiarizedSpeaker} and identity {IdentityId}: {SegmentCount} segment(s), {KnownSnippetCount} known snippet(s)",
request.DiarizedSpeaker,
request.IdentityId,
segments.Count,
layout.KnownSegments.Count);
var unknownSpeaker = FindBestSpeaker(layout.UnknownSegment.Value, segments);
if (string.IsNullOrWhiteSpace(unknownSpeaker))
{
@@ -98,11 +123,22 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
if (!accepted)
{
logger.LogInformation(
"pyannote rejected speaker identity match for {DiarizedSpeaker} and identity {IdentityId}: {Matching}/{Compared} known sample(s) matched",
"pyannote rejected speaker identity match for {DiarizedSpeaker} and identity {IdentityId}: {Matching}/{Compared} known sample(s) matched, required ratio {RequiredRatio:P2}",
request.DiarizedSpeaker,
request.IdentityId,
matching,
compared);
compared,
minimumRatio);
}
else
{
logger.LogInformation(
"pyannote accepted speaker identity match for {DiarizedSpeaker} and identity {IdentityId}: {Matching}/{Compared} known sample(s) matched, required ratio {RequiredRatio:P2}",
request.DiarizedSpeaker,
request.IdentityId,
matching,
compared,
minimumRatio);
}
return accepted;
@@ -162,7 +198,7 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
cancellationToken);
}
private bool HasSingleSpeaker(IReadOnlyList<TranscriptionSegment> segments)
private SingleSpeakerCoverage AnalyzeSingleSpeakerCoverage(IReadOnlyList<TranscriptionSegment> segments)
{
var durationsBySpeaker = segments
.Where(segment => !string.IsNullOrWhiteSpace(segment.Speaker) && segment.End > segment.Start)
@@ -176,17 +212,29 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
.ToList();
if (durationsBySpeaker.Count == 0)
{
return false;
}
if (durationsBySpeaker.Count == 1)
{
return true;
return new SingleSpeakerCoverage(false, 0, null, 0, TimeSpan.Zero);
}
var total = durationsBySpeaker.Sum(group => group.Duration);
var dominant = durationsBySpeaker.Max(group => group.Duration);
return total > 0 && dominant / total >= Math.Clamp(options.MinimumSingleSpeakerCoverage, 0, 1);
var dominant = durationsBySpeaker.OrderByDescending(group => group.Duration).First();
var coverage = total > 0 ? dominant.Duration / total : 0;
if (durationsBySpeaker.Count == 1)
{
return new SingleSpeakerCoverage(
true,
durationsBySpeaker.Count,
dominant.Speaker,
coverage,
TimeSpan.FromSeconds(total));
}
var accepted = total > 0 && coverage >= Math.Clamp(options.MinimumSingleSpeakerCoverage, 0, 1);
return new SingleSpeakerCoverage(
accepted,
durationsBySpeaker.Count,
dominant.Speaker,
coverage,
TimeSpan.FromSeconds(total));
}
private CompositeLayout WriteCompositeWav(
@@ -234,4 +282,11 @@ public sealed class PyannoteSpeakerIdentityMatchValidator : ISpeakerIdentityMatc
}
private sealed record CompositeLayout(TimeSegment? UnknownSegment, IReadOnlyList<TimeSegment> KnownSegments);
private sealed record SingleSpeakerCoverage(
bool Accepted,
int SpeakerCount,
string? DominantSpeaker,
double Coverage,
TimeSpan TotalDuration);
}