feat: add local Resemblyzer speaker recognition

2026-09-11 13:49:27 +02:00
parent 43fc8aaec0
commit f86af983e8
48 changed files with 5250 additions and 273 deletions
@@ -2899,6 +2899,65 @@ public sealed class RecordingCoordinatorTests
        Assert.Equal(["Chris"], speakerIdentification.Requests.Last().MeetingNote.Frontmatter.Attendees);
    }

+    [Fact]
+    public async Task LiveResemblyzerMatchingRetriesWhenSameSpeakerCollectsRequiredSamples()
+    {
+        var audioSource = new ControlledAudioSource();
+        var transcriptStore = new InMemoryTranscriptStore();
+        var speakerIdentification = new CountingSpeakerIdentificationService();
+        var coordinator = new MeetingRecordingCoordinator(
+            audioSource,
+            new TestSpeechRecognitionPipelineFactory(new OrderedChunkProvider()),
+            transcriptStore,
+            new InMemoryMeetingNoteStore(),
+            new CapturingMeetingNoteOpener(),
+            new InMemoryMeetingArtifactStore(),
+            new InMemoryRecordedAudioStore(),
+            new CapturingMeetingSummaryPipeline(),
+            Options.Create(new MeetingAssistantOptions
+            {
+                SpeakerIdentification = new SpeakerIdentificationOptions
+                {
+                    InitialDelay = TimeSpan.Zero,
+                    Interval = TimeSpan.FromMilliseconds(20),
+                    MinimumSampleSpeechDuration = TimeSpan.Zero,
+                    Resemblyzer = new ResemblyzerSpeakerRecognitionOptions
+                    {
+                        Enabled = true,
+                        RequiredVectorsPerSpeaker = 5
+                    }
+                }
+            }),
+            NullLogger<MeetingRecordingCoordinator>.Instance,
+            speakerIdentificationService: speakerIdentification,
+            speakerSampleCollectionPolicy: SpeakerSampleCollectionPolicy.IndependentVectors(1000));
+
+        await coordinator.StartAsync(CancellationToken.None);
+        try
+        {
+            await audioSource.WriteAsync(new AudioChunk(Samples(0, 1, 2, 3, 4, 5, 6, 7), 4, 1), CancellationToken.None);
+            await WaitUntilAsync(() => speakerIdentification.Requests.Count == 1);
+            Assert.Single(speakerIdentification.Requests.Single().Samples!);
+
+            // The provider emits overlapping two-second segments, so every second
+            // segment supplies another independent recognition sample.
+            for (var index = 1; index < 9; index++)
+            {
+                await audioSource.WriteAsync(new AudioChunk(Samples(0, 1, 2, 3, 4, 5, 6, 7), 4, 1), CancellationToken.None);
+            }
+
+            await WaitUntilAsync(() => speakerIdentification.Requests.Last().Samples?.Count == 5);
+            var attempts = speakerIdentification.Requests.Count;
+            await Task.Delay(100);
+            Assert.Equal(attempts, speakerIdentification.Requests.Count);
+            Assert.All(speakerIdentification.Requests.Last().Samples!, sample => Assert.Equal("Guest03", sample.Speaker));
+        }
+        finally
+        {
+            await coordinator.StopAsync(CancellationToken.None);
+        }
+    }
+
    [Fact]
    public async Task LiveSpeakerIdentificationRetriesWhenNewUnmappedSpeakerAppears()
    {