Public Access
Add inactivity safeguard and speaker diagnostics
PR and Push Build/Test / build-and-test (push) Failing after 8m31s
PR and Push Build/Test / build-and-test (push) Failing after 8m31s
This commit is contained in:
@@ -0,0 +1,18 @@
|
||||
namespace MeetingAssistant.Recording;
|
||||
|
||||
public interface IMeetingInactivityClock
|
||||
{
|
||||
DateTimeOffset Now { get; }
|
||||
|
||||
Task DelayAsync(TimeSpan delay, CancellationToken cancellationToken);
|
||||
}
|
||||
|
||||
public sealed class SystemMeetingInactivityClock : IMeetingInactivityClock
|
||||
{
|
||||
public DateTimeOffset Now => DateTimeOffset.Now;
|
||||
|
||||
public Task DelayAsync(TimeSpan delay, CancellationToken cancellationToken)
|
||||
{
|
||||
return Task.Delay(delay, cancellationToken);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,33 @@
|
||||
namespace MeetingAssistant.Recording;
|
||||
|
||||
public interface IMeetingInactivityPromptService
|
||||
{
|
||||
Task ShowStopPromptAsync(
|
||||
MeetingInactivityPromptRequest request,
|
||||
Func<MeetingInactivityPromptResponse, CancellationToken, Task> handleResponseAsync,
|
||||
CancellationToken cancellationToken);
|
||||
}
|
||||
|
||||
public sealed record MeetingInactivityPromptRequest(
|
||||
TimeSpan InactivityDuration,
|
||||
DateTimeOffset SuggestedEndTime,
|
||||
int PromptIndex,
|
||||
TimeSpan Threshold);
|
||||
|
||||
public enum MeetingInactivityPromptResponse
|
||||
{
|
||||
Dismissed,
|
||||
Continue,
|
||||
Stop
|
||||
}
|
||||
|
||||
public sealed class NoopMeetingInactivityPromptService : IMeetingInactivityPromptService
|
||||
{
|
||||
public Task ShowStopPromptAsync(
|
||||
MeetingInactivityPromptRequest request,
|
||||
Func<MeetingInactivityPromptResponse, CancellationToken, Task> handleResponseAsync,
|
||||
CancellationToken cancellationToken)
|
||||
{
|
||||
return Task.CompletedTask;
|
||||
}
|
||||
}
|
||||
@@ -29,6 +29,8 @@ public sealed class MeetingRecordingCoordinator
|
||||
private readonly IMeetingWorkflowEngine meetingWorkflowEngine;
|
||||
private readonly IMeetingScreenshotService screenshotService;
|
||||
private readonly IMeetingRunArtifactCleaner artifactCleaner;
|
||||
private readonly IMeetingInactivityPromptService inactivityPromptService;
|
||||
private readonly IMeetingInactivityClock inactivityClock;
|
||||
private readonly MeetingAssistantOptions options;
|
||||
private readonly ILogger<MeetingRecordingCoordinator> logger;
|
||||
private readonly SemaphoreSlim gate = new(1, 1);
|
||||
@@ -55,7 +57,9 @@ public sealed class MeetingRecordingCoordinator
|
||||
ILaunchProfileOptionsProvider? launchProfiles = null,
|
||||
IMeetingWorkflowEngine? meetingWorkflowEngine = null,
|
||||
IMeetingScreenshotService? screenshotService = null,
|
||||
IMeetingRunArtifactCleaner? artifactCleaner = null)
|
||||
IMeetingRunArtifactCleaner? artifactCleaner = null,
|
||||
IMeetingInactivityPromptService? inactivityPromptService = null,
|
||||
IMeetingInactivityClock? inactivityClock = null)
|
||||
{
|
||||
this.audioSource = audioSource;
|
||||
this.speechRecognitionPipelineFactory = speechRecognitionPipelineFactory;
|
||||
@@ -73,6 +77,8 @@ public sealed class MeetingRecordingCoordinator
|
||||
this.meetingWorkflowEngine = meetingWorkflowEngine ?? NoopMeetingWorkflowEngine.Instance;
|
||||
this.screenshotService = screenshotService ?? NoopMeetingScreenshotService.Instance;
|
||||
this.artifactCleaner = artifactCleaner ?? new MeetingRunArtifactCleaner();
|
||||
this.inactivityPromptService = inactivityPromptService ?? new NoopMeetingInactivityPromptService();
|
||||
this.inactivityClock = inactivityClock ?? new SystemMeetingInactivityClock();
|
||||
this.options = options.Value;
|
||||
this.logger = logger;
|
||||
}
|
||||
@@ -152,7 +158,7 @@ public sealed class MeetingRecordingCoordinator
|
||||
var runOptions = launchProfile.Options;
|
||||
currentSession = await transcriptStore.CreateSessionAsync(runOptions, cancellationToken);
|
||||
var recordedAudio = await recordedAudioStore.CreateSessionAsync(runOptions, cancellationToken);
|
||||
var startedAt = DateTimeOffset.Now;
|
||||
var startedAt = inactivityClock.Now;
|
||||
var assistantContextPath = GetAssistantContextPath(startedAt, runOptions);
|
||||
var summaryPath = GetSummaryPath(startedAt, runOptions);
|
||||
currentMeetingNote = await meetingNoteStore.SaveAsync(
|
||||
@@ -206,8 +212,16 @@ public sealed class MeetingRecordingCoordinator
|
||||
startedAt,
|
||||
launchProfile.Name,
|
||||
runOptions.SpeakerIdentification.LiveSampleBufferDuration,
|
||||
runOptions.SpeakerIdentification.MaxSnippetsPerSpeaker);
|
||||
runOptions.SpeakerIdentification.MaxSnippetsPerSpeaker,
|
||||
logger);
|
||||
run.Task = Task.Run(() => RecordAsync(run), CancellationToken.None);
|
||||
if (runOptions.Recording.InactivitySafeguard.Enabled)
|
||||
{
|
||||
_ = Task.Run(
|
||||
() => RunInactivitySafeguardAsync(run),
|
||||
CancellationToken.None);
|
||||
}
|
||||
|
||||
currentRun = run;
|
||||
_ = Task.Run(
|
||||
() => ApplyMeetingMetadataWhenAvailableAsync(startedAt, run),
|
||||
@@ -231,6 +245,13 @@ public sealed class MeetingRecordingCoordinator
|
||||
}
|
||||
|
||||
public async Task<RecordingStatus> StopAsync(CancellationToken cancellationToken)
|
||||
{
|
||||
return await StopAsync(null, cancellationToken);
|
||||
}
|
||||
|
||||
private async Task<RecordingStatus> StopAsync(
|
||||
DateTimeOffset? inferredEndTime,
|
||||
CancellationToken cancellationToken)
|
||||
{
|
||||
RecordingRun run;
|
||||
var abortAsTooShort = false;
|
||||
@@ -244,6 +265,11 @@ public sealed class MeetingRecordingCoordinator
|
||||
}
|
||||
|
||||
run = currentRun;
|
||||
if (inferredEndTime is not null)
|
||||
{
|
||||
run.SetInferredEndTime(inferredEndTime.Value);
|
||||
}
|
||||
|
||||
abortAsTooShort = ShouldAbortRunOnStop(run, DateTimeOffset.Now);
|
||||
if (abortAsTooShort)
|
||||
{
|
||||
@@ -461,6 +487,7 @@ public sealed class MeetingRecordingCoordinator
|
||||
var completedMeetingNote = await CompleteMeetingNoteAsync(
|
||||
run.MeetingNotePath,
|
||||
run.Options,
|
||||
run.InferredEndTime,
|
||||
run.TranscriptionCancellation);
|
||||
if (completedMeetingNote is not null)
|
||||
{
|
||||
@@ -547,13 +574,121 @@ public sealed class MeetingRecordingCoordinator
|
||||
{
|
||||
await foreach (var segment in pipeline.ReadLiveTranscriptAsync(cancellationToken))
|
||||
{
|
||||
run.TryAddSpeakerSample(segment);
|
||||
run.RecordTranscriptActivity(segment, inactivityClock.Now);
|
||||
var sample = run.TryAddSpeakerSample(segment);
|
||||
if (sample is not null)
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Collected speaker identity sample for {Speaker}: start {Start}, end {End}, duration {Duration}, score {Score}, sample bytes {SampleBytes}",
|
||||
sample.Speaker,
|
||||
sample.Segment.Start,
|
||||
sample.Segment.End,
|
||||
sample.Segment.End - sample.Segment.Start,
|
||||
sample.Score,
|
||||
sample.WavBytes.Length);
|
||||
}
|
||||
|
||||
var relabeledSegment = run.Relabel(segment);
|
||||
run.AddLiveSegment(relabeledSegment);
|
||||
await transcriptStore.AppendAsync(run.Session, relabeledSegment, cancellationToken);
|
||||
}
|
||||
}
|
||||
|
||||
private async Task RunInactivitySafeguardAsync(RecordingRun run)
|
||||
{
|
||||
var safeguardOptions = run.Options.Recording.InactivitySafeguard;
|
||||
var promptThresholds = GetInactivityPromptThresholds(safeguardOptions);
|
||||
var promptedThresholds = new HashSet<TimeSpan>();
|
||||
var lastActivityVersion = run.GetTranscriptActivitySnapshot().ActivityVersion;
|
||||
|
||||
logger.LogInformation(
|
||||
"Recording inactivity safeguard started with prompt thresholds {PromptThresholds}, auto-stop {AutoStopAfter}, check interval {CheckInterval}",
|
||||
string.Join(", ", promptThresholds),
|
||||
safeguardOptions.AutoStopAfter,
|
||||
safeguardOptions.CheckInterval);
|
||||
|
||||
try
|
||||
{
|
||||
while (!run.CaptureCancellation.IsCancellationRequested && !run.IsAborted)
|
||||
{
|
||||
var snapshot = run.GetTranscriptActivitySnapshot();
|
||||
if (snapshot.ActivityVersion != lastActivityVersion)
|
||||
{
|
||||
promptedThresholds.Clear();
|
||||
lastActivityVersion = snapshot.ActivityVersion;
|
||||
logger.LogInformation(
|
||||
"Recording inactivity safeguard reset after transcript activity at {LastTranscriptActivityAt}",
|
||||
snapshot.LastTranscriptActivityAt);
|
||||
}
|
||||
|
||||
var now = inactivityClock.Now;
|
||||
var inactivityDuration = now - snapshot.LastTranscriptActivityAt;
|
||||
if (safeguardOptions.AutoStopAfter > TimeSpan.Zero &&
|
||||
inactivityDuration >= safeguardOptions.AutoStopAfter)
|
||||
{
|
||||
var inferredEndTime = run.GetInactivitySafeguardEndTime(safeguardOptions.InferredEndPadding);
|
||||
logger.LogWarning(
|
||||
"Recording inactivity safeguard auto-stopping meeting after {InactivityDuration} without transcript text; inferred end time {InferredEndTime}",
|
||||
inactivityDuration,
|
||||
inferredEndTime);
|
||||
await StopAsync(inferredEndTime, CancellationToken.None);
|
||||
return;
|
||||
}
|
||||
|
||||
foreach (var threshold in promptThresholds)
|
||||
{
|
||||
if (promptedThresholds.Contains(threshold) || inactivityDuration < threshold)
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
promptedThresholds.Add(threshold);
|
||||
var inferredEndTime = run.GetInactivitySafeguardEndTime(safeguardOptions.InferredEndPadding);
|
||||
logger.LogInformation(
|
||||
"Recording inactivity safeguard prompting after {InactivityDuration} without transcript text at threshold {Threshold}; inferred end time {InferredEndTime}",
|
||||
inactivityDuration,
|
||||
threshold,
|
||||
inferredEndTime);
|
||||
await inactivityPromptService.ShowStopPromptAsync(
|
||||
new MeetingInactivityPromptRequest(
|
||||
inactivityDuration,
|
||||
inferredEndTime,
|
||||
promptedThresholds.Count,
|
||||
threshold),
|
||||
async (response, cancellationToken) =>
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Recording inactivity safeguard prompt response {Response} at threshold {Threshold}",
|
||||
response,
|
||||
threshold);
|
||||
|
||||
if (response == MeetingInactivityPromptResponse.Stop)
|
||||
{
|
||||
await StopAsync(inferredEndTime, CancellationToken.None);
|
||||
}
|
||||
},
|
||||
run.CaptureCancellation);
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
await inactivityClock.DelayAsync(
|
||||
GetInactivitySafeguardCheckInterval(safeguardOptions),
|
||||
run.CaptureCancellation);
|
||||
}
|
||||
}
|
||||
catch (OperationCanceledException) when (run.CaptureCancellation.IsCancellationRequested)
|
||||
{
|
||||
}
|
||||
catch (ObjectDisposedException) when (run.CaptureCancellation.IsCancellationRequested)
|
||||
{
|
||||
}
|
||||
catch (Exception exception)
|
||||
{
|
||||
logger.LogError(exception, "Recording inactivity safeguard failed");
|
||||
}
|
||||
}
|
||||
|
||||
private async Task IdentifyLiveSpeakersAsync(
|
||||
RecordingRun run,
|
||||
CancellationToken cancellationToken)
|
||||
@@ -596,6 +731,10 @@ public sealed class MeetingRecordingCoordinator
|
||||
var samples = run.GetSpeakerSamplesSnapshot();
|
||||
if (segments.Count == 0 || samples.Count == 0)
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Skipping live speaker identity matching: {SegmentCount} live segment(s), {SampleCount} collected sample(s)",
|
||||
segments.Count,
|
||||
samples.Count);
|
||||
return;
|
||||
}
|
||||
|
||||
@@ -605,9 +744,15 @@ public sealed class MeetingRecordingCoordinator
|
||||
var checkpoint = run.CreateLiveIdentificationCheckpoint(meetingNote);
|
||||
if (checkpoint is null)
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Skipping live speaker identity matching because no new unmapped sample speakers or attendee changes were found");
|
||||
return;
|
||||
}
|
||||
|
||||
logger.LogInformation(
|
||||
"Running live speaker identity matching for speakers {Speakers} with {SampleCount} sample(s)",
|
||||
string.Join(", ", checkpoint.Speakers),
|
||||
samples.Count);
|
||||
var result = await speakerIdentificationService!.IdentifyKnownSpeakersAsync(
|
||||
new SpeakerIdentificationRequest(
|
||||
"",
|
||||
@@ -830,11 +975,18 @@ public sealed class MeetingRecordingCoordinator
|
||||
if (run.HasSwitchedProfile)
|
||||
{
|
||||
var liveSegments = run.GetLiveSegmentsSnapshot();
|
||||
var liveSamples = run.GetSpeakerSamplesSnapshot();
|
||||
var liveMappings = run.GetSpeakerMappingsSnapshot();
|
||||
run.SetPendingFinalSpeakerIdentification(
|
||||
run.RecordedAudio.AudioPath,
|
||||
liveSegments,
|
||||
run.GetSpeakerSamplesSnapshot(),
|
||||
run.GetSpeakerMappingsSnapshot());
|
||||
liveSamples,
|
||||
liveMappings);
|
||||
logger.LogInformation(
|
||||
"Queued pending final speaker identity processing after profile switch: {SegmentCount} segment(s), {SampleCount} sample(s), {KnownMappingCount} known mapping(s)",
|
||||
liveSegments.Count,
|
||||
liveSamples.Count,
|
||||
liveMappings.Count);
|
||||
await transcriptStore.ReplaceAsync(run.Session, liveSegments, cancellationToken);
|
||||
return;
|
||||
}
|
||||
@@ -875,6 +1027,11 @@ public sealed class MeetingRecordingCoordinator
|
||||
finishedSpeakerResult.Segments,
|
||||
samples,
|
||||
knownSpeakerMappings);
|
||||
logger.LogInformation(
|
||||
"Queued pending final speaker identity processing: {SegmentCount} segment(s), {SampleCount} sample(s), {KnownMappingCount} known mapping(s)",
|
||||
finishedSpeakerResult.Segments.Count,
|
||||
samples.Count,
|
||||
knownSpeakerMappings.Count);
|
||||
await transcriptStore.ReplaceAsync(run.Session, finishedSpeakerResult.Segments, cancellationToken);
|
||||
}
|
||||
|
||||
@@ -937,6 +1094,11 @@ public sealed class MeetingRecordingCoordinator
|
||||
|
||||
try
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Starting pending final speaker identity processing: {SegmentCount} segment(s), {SampleCount} sample(s), {KnownMappingCount} known mapping(s)",
|
||||
pending.Segments.Count,
|
||||
pending.Samples.Count,
|
||||
pending.KnownSpeakerMappings.Count);
|
||||
var meetingNote = await meetingNoteStore.ReadAsync(run.MeetingNotePath, cancellationToken);
|
||||
var segments = pending.Segments.ToList();
|
||||
var knownSpeakerMappings = pending.KnownSpeakerMappings.ToDictionary(
|
||||
@@ -948,6 +1110,10 @@ public sealed class MeetingRecordingCoordinator
|
||||
cancellationToken);
|
||||
foreach (var speakerOverride in speakerOverrides)
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Applying summary speaker override before final identity learning: {SourceSpeaker} -> {TargetSpeaker}",
|
||||
speakerOverride.From,
|
||||
speakerOverride.To);
|
||||
await speakerIdentificationService.ApplySpeakerOverrideAsync(
|
||||
new SpeakerIdentificationRequest(
|
||||
pending.AudioPath,
|
||||
@@ -971,6 +1137,10 @@ public sealed class MeetingRecordingCoordinator
|
||||
cancellationToken);
|
||||
foreach (var deletion in identityDeletions)
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Applying summary speaker identity deletion before final identity learning: {Identity} -> {Replacement}",
|
||||
deletion.Identity,
|
||||
deletion.Replacement);
|
||||
await speakerIdentificationService.DeleteSpeakerIdentityAsync(
|
||||
deletion.Identity,
|
||||
cancellationToken);
|
||||
@@ -989,6 +1159,10 @@ public sealed class MeetingRecordingCoordinator
|
||||
pending.Samples,
|
||||
knownSpeakerMappings),
|
||||
cancellationToken);
|
||||
logger.LogInformation(
|
||||
"Completed final speaker identity learning: {MappingCount} mapping(s), {SegmentCount} segment(s)",
|
||||
result.SpeakerMappings.Count,
|
||||
result.Segments.Count);
|
||||
await AddIdentifiedSpeakersToMeetingAttendeesAsync(
|
||||
result.AttendeeMatches,
|
||||
run.Artifacts,
|
||||
@@ -1089,6 +1263,7 @@ public sealed class MeetingRecordingCoordinator
|
||||
private async Task<MeetingNote?> CompleteMeetingNoteAsync(
|
||||
string meetingNotePath,
|
||||
MeetingAssistantOptions runOptions,
|
||||
DateTimeOffset? endTime,
|
||||
CancellationToken cancellationToken)
|
||||
{
|
||||
if (string.IsNullOrWhiteSpace(meetingNotePath))
|
||||
@@ -1097,7 +1272,7 @@ public sealed class MeetingRecordingCoordinator
|
||||
}
|
||||
|
||||
var meetingNote = await meetingNoteStore.ReadAsync(meetingNotePath, cancellationToken);
|
||||
meetingNote.Frontmatter.EndTime = DateTimeOffset.Now;
|
||||
meetingNote.Frontmatter.EndTime = endTime ?? DateTimeOffset.Now;
|
||||
var savedMeetingNote = await meetingNoteStore.SaveAsync(meetingNote, runOptions, cancellationToken);
|
||||
if (string.Equals(currentMeetingNote?.Path, savedMeetingNote.Path, StringComparison.OrdinalIgnoreCase))
|
||||
{
|
||||
@@ -1294,6 +1469,26 @@ public sealed class MeetingRecordingCoordinator
|
||||
return minimumDuration > TimeSpan.Zero && stoppedAt - run.StartedAt < minimumDuration;
|
||||
}
|
||||
|
||||
private static IReadOnlyList<TimeSpan> GetInactivityPromptThresholds(
|
||||
RecordingInactivitySafeguardOptions options)
|
||||
{
|
||||
return new[] { options.FirstPromptAfter }
|
||||
.Concat(options.ReminderPromptAfter ?? [])
|
||||
.Where(threshold => threshold > TimeSpan.Zero)
|
||||
.Where(threshold => options.AutoStopAfter <= TimeSpan.Zero || threshold < options.AutoStopAfter)
|
||||
.Distinct()
|
||||
.Order()
|
||||
.ToList();
|
||||
}
|
||||
|
||||
private static TimeSpan GetInactivitySafeguardCheckInterval(
|
||||
RecordingInactivitySafeguardOptions options)
|
||||
{
|
||||
return options.CheckInterval > TimeSpan.Zero
|
||||
? options.CheckInterval
|
||||
: TimeSpan.FromSeconds(15);
|
||||
}
|
||||
|
||||
private async Task CompleteRunAsync(RecordingRun run)
|
||||
{
|
||||
if (!run.TryComplete())
|
||||
@@ -1326,6 +1521,7 @@ public sealed class MeetingRecordingCoordinator
|
||||
private readonly object liveSegmentsGate = new();
|
||||
private readonly object liveTranscriptTasksGate = new();
|
||||
private readonly object liveIdentificationGate = new();
|
||||
private readonly object transcriptActivityGate = new();
|
||||
private readonly List<TranscriptionSegment> liveSegments = [];
|
||||
private readonly List<Task> liveTranscriptTasks = [];
|
||||
private readonly List<ISpeechRecognitionPipeline> pipelines = [];
|
||||
@@ -1333,6 +1529,10 @@ public sealed class MeetingRecordingCoordinator
|
||||
private readonly SpeakerAudioSampleCollector speakerSampleCollector;
|
||||
private LiveIdentificationCheckpoint? lastLiveIdentificationCheckpoint;
|
||||
private FinalSpeakerIdentificationWork? pendingFinalSpeakerIdentification;
|
||||
private DateTimeOffset lastTranscriptActivityAt;
|
||||
private TimeSpan? lastTranscriptSegmentEnd;
|
||||
private long transcriptActivityVersion;
|
||||
private DateTimeOffset? inferredEndTime;
|
||||
|
||||
public RecordingRun(
|
||||
CancellationTokenSource captureCancellation,
|
||||
@@ -1347,7 +1547,8 @@ public sealed class MeetingRecordingCoordinator
|
||||
DateTimeOffset startedAt,
|
||||
string launchProfileName,
|
||||
TimeSpan liveSampleBufferDuration,
|
||||
int maxSpeakerSamples)
|
||||
int maxSpeakerSamples,
|
||||
ILogger logger)
|
||||
{
|
||||
CaptureCancellationSource = captureCancellation;
|
||||
TranscriptionCancellationSource = transcriptionCancellation;
|
||||
@@ -1360,13 +1561,15 @@ public sealed class MeetingRecordingCoordinator
|
||||
PipelineOptions = pipelineOptions;
|
||||
Options = options;
|
||||
StartedAt = startedAt;
|
||||
lastTranscriptActivityAt = startedAt;
|
||||
LaunchProfileName = launchProfileName;
|
||||
pipelines.Add(pipeline);
|
||||
speakerSampleCollector = new SpeakerAudioSampleCollector(
|
||||
liveSampleBufferDuration,
|
||||
maxSpeakerSamples,
|
||||
options.SpeakerIdentification.MinimumSampleSpeechDuration,
|
||||
options.SpeakerIdentification.MaximumSampleSegmentGap);
|
||||
options.SpeakerIdentification.MaximumSampleSegmentGap,
|
||||
logger);
|
||||
}
|
||||
|
||||
public CancellationTokenSource CaptureCancellationSource { get; }
|
||||
@@ -1411,6 +1614,17 @@ public sealed class MeetingRecordingCoordinator
|
||||
|
||||
public AssistantContextState ContextState { get; private set; } = AssistantContextState.CollectingMetadata;
|
||||
|
||||
public DateTimeOffset? InferredEndTime
|
||||
{
|
||||
get
|
||||
{
|
||||
lock (transcriptActivityGate)
|
||||
{
|
||||
return inferredEndTime;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
public void StopCapture()
|
||||
{
|
||||
CaptureCancellationSource.Cancel();
|
||||
@@ -1557,14 +1771,59 @@ public sealed class MeetingRecordingCoordinator
|
||||
}
|
||||
}
|
||||
|
||||
public void RecordTranscriptActivity(TranscriptionSegment segment, DateTimeOffset arrivedAt)
|
||||
{
|
||||
if (string.IsNullOrWhiteSpace(segment.Text))
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
lock (transcriptActivityGate)
|
||||
{
|
||||
lastTranscriptActivityAt = arrivedAt;
|
||||
var segmentEnd = segment.End > TimeSpan.Zero ? segment.End : segment.Start;
|
||||
lastTranscriptSegmentEnd = lastTranscriptSegmentEnd is null
|
||||
? segmentEnd
|
||||
: TimeSpan.FromTicks(Math.Max(lastTranscriptSegmentEnd.Value.Ticks, segmentEnd.Ticks));
|
||||
transcriptActivityVersion++;
|
||||
}
|
||||
}
|
||||
|
||||
public TranscriptActivitySnapshot GetTranscriptActivitySnapshot()
|
||||
{
|
||||
lock (transcriptActivityGate)
|
||||
{
|
||||
return new TranscriptActivitySnapshot(
|
||||
lastTranscriptActivityAt,
|
||||
lastTranscriptSegmentEnd,
|
||||
transcriptActivityVersion);
|
||||
}
|
||||
}
|
||||
|
||||
public DateTimeOffset GetInactivitySafeguardEndTime(TimeSpan padding)
|
||||
{
|
||||
lock (transcriptActivityGate)
|
||||
{
|
||||
return StartedAt + (lastTranscriptSegmentEnd ?? TimeSpan.Zero) + padding;
|
||||
}
|
||||
}
|
||||
|
||||
public void SetInferredEndTime(DateTimeOffset endTime)
|
||||
{
|
||||
lock (transcriptActivityGate)
|
||||
{
|
||||
inferredEndTime = endTime;
|
||||
}
|
||||
}
|
||||
|
||||
public void AppendAudio(AudioChunk chunk)
|
||||
{
|
||||
speakerSampleCollector.AppendAudio(chunk);
|
||||
}
|
||||
|
||||
public void TryAddSpeakerSample(TranscriptionSegment segment)
|
||||
public SpeakerAudioSample? TryAddSpeakerSample(TranscriptionSegment segment)
|
||||
{
|
||||
speakerSampleCollector.TryAdd(segment);
|
||||
return speakerSampleCollector.TryAdd(segment);
|
||||
}
|
||||
|
||||
public IReadOnlyList<TranscriptionSegment> GetLiveSegmentsSnapshot()
|
||||
@@ -1683,6 +1942,11 @@ public sealed class MeetingRecordingCoordinator
|
||||
PipelineGate.Dispose();
|
||||
}
|
||||
|
||||
public sealed record TranscriptActivitySnapshot(
|
||||
DateTimeOffset LastTranscriptActivityAt,
|
||||
TimeSpan? LastTranscriptSegmentEnd,
|
||||
long ActivityVersion);
|
||||
|
||||
private static int StateRank(AssistantContextState state)
|
||||
{
|
||||
return state switch
|
||||
|
||||
@@ -1,5 +1,6 @@
|
||||
using MeetingAssistant.Speakers;
|
||||
using MeetingAssistant.Transcription;
|
||||
using Microsoft.Extensions.Logging;
|
||||
|
||||
namespace MeetingAssistant.Recording;
|
||||
|
||||
@@ -11,6 +12,7 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
private readonly int maxSamplesPerSpeaker;
|
||||
private readonly TimeSpan minimumUninterruptedSpeechDuration;
|
||||
private readonly TimeSpan maximumSegmentGap;
|
||||
private readonly ILogger? logger;
|
||||
private PendingSpeakerSpan? pendingSpan;
|
||||
|
||||
public SpeakerAudioSampleCollector(TimeSpan bufferDuration, int maxSamplesPerSpeaker)
|
||||
@@ -18,7 +20,8 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
bufferDuration,
|
||||
maxSamplesPerSpeaker,
|
||||
TimeSpan.FromSeconds(30),
|
||||
TimeSpan.FromSeconds(1))
|
||||
TimeSpan.FromSeconds(1),
|
||||
logger: null)
|
||||
{
|
||||
}
|
||||
|
||||
@@ -26,7 +29,8 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
TimeSpan bufferDuration,
|
||||
int maxSamplesPerSpeaker,
|
||||
TimeSpan minimumUninterruptedSpeechDuration,
|
||||
TimeSpan maximumSegmentGap)
|
||||
TimeSpan maximumSegmentGap,
|
||||
ILogger? logger = null)
|
||||
{
|
||||
audioBuffer = new RollingAudioBuffer(bufferDuration);
|
||||
this.maxSamplesPerSpeaker = Math.Max(1, maxSamplesPerSpeaker);
|
||||
@@ -36,6 +40,7 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
this.maximumSegmentGap = maximumSegmentGap >= TimeSpan.Zero
|
||||
? maximumSegmentGap
|
||||
: TimeSpan.Zero;
|
||||
this.logger = logger;
|
||||
}
|
||||
|
||||
public void AppendAudio(AudioChunk chunk)
|
||||
@@ -53,32 +58,61 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
}
|
||||
}
|
||||
|
||||
public void TryAdd(TranscriptionSegment segment)
|
||||
public SpeakerAudioSample? TryAdd(TranscriptionSegment segment)
|
||||
{
|
||||
if (!IsDiarizedSpeaker(segment.Speaker))
|
||||
{
|
||||
return;
|
||||
logger?.LogInformation(
|
||||
"Discarding speaker identity sample for {Speaker} because the segment has no diarized speaker label",
|
||||
segment.Speaker);
|
||||
return null;
|
||||
}
|
||||
|
||||
TranscriptionSegment sampleSegment;
|
||||
PendingSpanReset? reset;
|
||||
lock (gate)
|
||||
{
|
||||
sampleSegment = ExtendPendingSpan(segment);
|
||||
(sampleSegment, reset) = ExtendPendingSpan(segment);
|
||||
}
|
||||
|
||||
if (reset is not null)
|
||||
{
|
||||
logger?.LogInformation(
|
||||
"Reset speaker identity sample span from {PreviousSpeaker} to {Speaker}: previous end {PreviousEnd}, next start {NextStart}, gap {Gap}, maximum gap {MaximumGap}",
|
||||
reset.PreviousSpeaker,
|
||||
segment.Speaker,
|
||||
reset.PreviousEnd,
|
||||
segment.Start,
|
||||
reset.Gap,
|
||||
maximumSegmentGap);
|
||||
}
|
||||
|
||||
var score = Score(sampleSegment, minimumUninterruptedSpeechDuration);
|
||||
if (score <= 0)
|
||||
if (!score.Accepted)
|
||||
{
|
||||
return;
|
||||
logger?.LogInformation(
|
||||
"Discarding speaker identity sample for {Speaker} because {Reason}: duration {Duration}, minimum duration {MinimumDuration}, word count {WordCount}",
|
||||
sampleSegment.Speaker,
|
||||
score.Reason,
|
||||
sampleSegment.End - sampleSegment.Start,
|
||||
minimumUninterruptedSpeechDuration,
|
||||
score.WordCount);
|
||||
return null;
|
||||
}
|
||||
|
||||
var wavBytes = audioBuffer.TryExtractWav(sampleSegment.Start, sampleSegment.End);
|
||||
if (wavBytes.Length == 0)
|
||||
{
|
||||
return;
|
||||
logger?.LogInformation(
|
||||
"Discarding speaker identity sample for {Speaker} because no audio could be extracted from rolling buffer: start {Start}, end {End}, duration {Duration}",
|
||||
sampleSegment.Speaker,
|
||||
sampleSegment.Start,
|
||||
sampleSegment.End,
|
||||
sampleSegment.End - sampleSegment.Start);
|
||||
return null;
|
||||
}
|
||||
|
||||
var sample = new SpeakerAudioSample(sampleSegment.Speaker, sampleSegment, wavBytes, score);
|
||||
var sample = new SpeakerAudioSample(sampleSegment.Speaker, sampleSegment, wavBytes, score.Value);
|
||||
lock (gate)
|
||||
{
|
||||
if (!samplesBySpeaker.TryGetValue(sampleSegment.Speaker, out var samples))
|
||||
@@ -88,13 +122,27 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
}
|
||||
|
||||
samples.Add(sample);
|
||||
var beforeCount = samples.Count;
|
||||
var bestSamples = samples
|
||||
.OrderByDescending(candidate => candidate.Score)
|
||||
.Take(maxSamplesPerSpeaker)
|
||||
.ToList();
|
||||
var retained = bestSamples.Contains(sample);
|
||||
samples.Clear();
|
||||
samples.AddRange(bestSamples);
|
||||
if (!retained)
|
||||
{
|
||||
logger?.LogInformation(
|
||||
"Discarding speaker identity sample for {Speaker} because it was not among the best {MaxSamplesPerSpeaker} retained sample(s): score {Score}, candidate count {CandidateCount}",
|
||||
sample.Speaker,
|
||||
maxSamplesPerSpeaker,
|
||||
sample.Score,
|
||||
beforeCount);
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
return sample;
|
||||
}
|
||||
|
||||
public IReadOnlyList<SpeakerAudioSample> Snapshot()
|
||||
@@ -115,39 +163,43 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
!string.Equals(speaker, "Unknown", StringComparison.OrdinalIgnoreCase);
|
||||
}
|
||||
|
||||
private TranscriptionSegment ExtendPendingSpan(TranscriptionSegment segment)
|
||||
private (TranscriptionSegment Segment, PendingSpanReset? Reset) ExtendPendingSpan(TranscriptionSegment segment)
|
||||
{
|
||||
if (pendingSpan is null ||
|
||||
!SpeakerSampleSpanSelector.CanExtend(pendingSpan.Speaker, pendingSpan.End, segment, maximumSegmentGap))
|
||||
{
|
||||
var reset = pendingSpan is null
|
||||
? null
|
||||
: new PendingSpanReset(
|
||||
pendingSpan.Speaker,
|
||||
pendingSpan.End,
|
||||
segment.Start - pendingSpan.End);
|
||||
pendingSpan = new PendingSpeakerSpan(
|
||||
segment.Speaker,
|
||||
segment.Start,
|
||||
segment.End,
|
||||
[segment.Text]);
|
||||
return pendingSpan.ToSegment();
|
||||
return (pendingSpan.ToSegment(), reset);
|
||||
}
|
||||
|
||||
pendingSpan = pendingSpan.Extend(segment);
|
||||
return pendingSpan.ToSegment();
|
||||
return (pendingSpan.ToSegment(), null);
|
||||
}
|
||||
|
||||
private static double Score(
|
||||
private static SampleScore Score(
|
||||
TranscriptionSegment segment,
|
||||
TimeSpan minimumUninterruptedSpeechDuration)
|
||||
{
|
||||
var durationSeconds = (segment.End - segment.Start).TotalSeconds;
|
||||
if (durationSeconds < minimumUninterruptedSpeechDuration.TotalSeconds)
|
||||
{
|
||||
return 0;
|
||||
return new SampleScore(false, 0, "speech duration is below the configured minimum", WordCount(segment.Text));
|
||||
}
|
||||
|
||||
var words = segment.Text
|
||||
.Split(' ', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries)
|
||||
.Length;
|
||||
var words = WordCount(segment.Text);
|
||||
if (words < 3)
|
||||
{
|
||||
return 0;
|
||||
return new SampleScore(false, 0, "word count is below the minimum useful sample length", words);
|
||||
}
|
||||
|
||||
var durationScore = Math.Min(durationSeconds / Math.Max(1, minimumUninterruptedSpeechDuration.TotalSeconds), 2);
|
||||
@@ -157,9 +209,20 @@ internal sealed class SpeakerAudioSampleCollector
|
||||
segment.Text.TrimEnd().EndsWith('!')
|
||||
? 5
|
||||
: 0;
|
||||
return durationScore * 70 + wordScore * 30 + sentenceBonus;
|
||||
return new SampleScore(true, durationScore * 70 + wordScore * 30 + sentenceBonus, null, words);
|
||||
}
|
||||
|
||||
private static int WordCount(string text)
|
||||
{
|
||||
return text
|
||||
.Split(' ', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries)
|
||||
.Length;
|
||||
}
|
||||
|
||||
private sealed record SampleScore(bool Accepted, double Value, string? Reason, int WordCount);
|
||||
|
||||
private sealed record PendingSpanReset(string PreviousSpeaker, TimeSpan PreviousEnd, TimeSpan Gap);
|
||||
|
||||
private sealed record PendingSpeakerSpan(
|
||||
string Speaker,
|
||||
TimeSpan Start,
|
||||
|
||||
@@ -0,0 +1,214 @@
|
||||
#if WINDOWS
|
||||
using System.Collections.Concurrent;
|
||||
using CommunityToolkit.WinUI.Notifications;
|
||||
|
||||
namespace MeetingAssistant.Recording;
|
||||
|
||||
public sealed class WindowsMeetingInactivityPromptService : IMeetingInactivityPromptService, IDisposable
|
||||
{
|
||||
private const string NotificationSource = "meeting-inactivity-safeguard";
|
||||
private const string NotificationGroup = "meeting-inactivity";
|
||||
private readonly ConcurrentDictionary<string, PendingPrompt> pendingPrompts = new(StringComparer.OrdinalIgnoreCase);
|
||||
private readonly ILogger<WindowsMeetingInactivityPromptService> logger;
|
||||
private readonly object registrationGate = new();
|
||||
private bool registered;
|
||||
|
||||
public WindowsMeetingInactivityPromptService(ILogger<WindowsMeetingInactivityPromptService> logger)
|
||||
{
|
||||
this.logger = logger;
|
||||
}
|
||||
|
||||
public Task ShowStopPromptAsync(
|
||||
MeetingInactivityPromptRequest request,
|
||||
Func<MeetingInactivityPromptResponse, CancellationToken, Task> handleResponseAsync,
|
||||
CancellationToken cancellationToken)
|
||||
{
|
||||
if (!OperatingSystem.IsWindowsVersionAtLeast(10, 0, 17763))
|
||||
{
|
||||
logger.LogWarning("Windows app notifications require Windows 10 version 1809 or later");
|
||||
return Task.CompletedTask;
|
||||
}
|
||||
|
||||
if (cancellationToken.IsCancellationRequested)
|
||||
{
|
||||
return Task.FromCanceled(cancellationToken);
|
||||
}
|
||||
|
||||
try
|
||||
{
|
||||
EnsureRegistered();
|
||||
var promptId = Guid.NewGuid().ToString("N");
|
||||
pendingPrompts[promptId] = new PendingPrompt(handleResponseAsync);
|
||||
var notification = BuildNotification(promptId, request);
|
||||
notification.Show(toast =>
|
||||
{
|
||||
toast.Group = NotificationGroup;
|
||||
toast.Tag = promptId;
|
||||
toast.ExpirationTime = DateTimeOffset.Now.AddMinutes(15);
|
||||
});
|
||||
logger.LogInformation(
|
||||
"Displayed native Windows inactivity notification {PromptId} at threshold {Threshold}",
|
||||
promptId,
|
||||
request.Threshold);
|
||||
}
|
||||
catch (Exception exception)
|
||||
{
|
||||
logger.LogError(exception, "Failed to display native Windows inactivity notification");
|
||||
}
|
||||
|
||||
return Task.CompletedTask;
|
||||
}
|
||||
|
||||
public void Dispose()
|
||||
{
|
||||
if (!registered)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
try
|
||||
{
|
||||
ToastNotificationManagerCompat.OnActivated -= OnNotificationInvoked;
|
||||
ToastNotificationManagerCompat.Uninstall();
|
||||
}
|
||||
catch (Exception exception)
|
||||
{
|
||||
logger.LogWarning(exception, "Failed to unregister native Windows toast notifications");
|
||||
}
|
||||
}
|
||||
|
||||
private void EnsureRegistered()
|
||||
{
|
||||
if (registered)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
lock (registrationGate)
|
||||
{
|
||||
if (registered)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
ToastNotificationManagerCompat.OnActivated += OnNotificationInvoked;
|
||||
registered = true;
|
||||
logger.LogInformation("Registered native Windows toast notification activation handler");
|
||||
}
|
||||
}
|
||||
|
||||
private static ToastContentBuilder BuildNotification(
|
||||
string promptId,
|
||||
MeetingInactivityPromptRequest request)
|
||||
{
|
||||
var yesButton = new ToastButton()
|
||||
.SetContent("Yes")
|
||||
.AddArgument("source", NotificationSource)
|
||||
.AddArgument("promptId", promptId)
|
||||
.AddArgument("response", "stop")
|
||||
.SetBackgroundActivation();
|
||||
|
||||
var noButton = new ToastButton()
|
||||
.SetContent("No")
|
||||
.AddArgument("source", NotificationSource)
|
||||
.AddArgument("promptId", promptId)
|
||||
.AddArgument("response", "continue")
|
||||
.SetBackgroundActivation();
|
||||
|
||||
return new ToastContentBuilder()
|
||||
.AddArgument("source", NotificationSource)
|
||||
.AddArgument("promptId", promptId)
|
||||
.AddText("Stop meeting?")
|
||||
.AddText($"No transcript text has arrived for {FormatDuration(request.InactivityDuration)}.")
|
||||
.AddButton(yesButton)
|
||||
.AddButton(noButton);
|
||||
}
|
||||
|
||||
private void OnNotificationInvoked(ToastNotificationActivatedEventArgsCompat args)
|
||||
{
|
||||
var arguments = ParseArguments(args.Argument);
|
||||
if (!TryGetArgument(arguments, "source", out var source) ||
|
||||
!string.Equals(source, NotificationSource, StringComparison.OrdinalIgnoreCase) ||
|
||||
!TryGetArgument(arguments, "promptId", out var promptId) ||
|
||||
!pendingPrompts.TryRemove(promptId, out var pendingPrompt))
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
var response = TryGetArgument(arguments, "response", out var responseValue) &&
|
||||
string.Equals(responseValue, "stop", StringComparison.OrdinalIgnoreCase)
|
||||
? MeetingInactivityPromptResponse.Stop
|
||||
: MeetingInactivityPromptResponse.Continue;
|
||||
_ = Task.Run(async () =>
|
||||
{
|
||||
try
|
||||
{
|
||||
logger.LogInformation(
|
||||
"Native Windows inactivity notification {PromptId} activated with response {Response}",
|
||||
promptId,
|
||||
response);
|
||||
await pendingPrompt.HandleResponseAsync(response, CancellationToken.None);
|
||||
}
|
||||
catch (Exception exception)
|
||||
{
|
||||
logger.LogError(
|
||||
exception,
|
||||
"Failed to handle native Windows inactivity notification response {Response}",
|
||||
response);
|
||||
}
|
||||
});
|
||||
}
|
||||
|
||||
private static IReadOnlyDictionary<string, string> ParseArguments(string argumentText)
|
||||
{
|
||||
if (string.IsNullOrWhiteSpace(argumentText))
|
||||
{
|
||||
return new Dictionary<string, string>();
|
||||
}
|
||||
|
||||
var arguments = new Dictionary<string, string>(StringComparer.OrdinalIgnoreCase);
|
||||
foreach (var pair in argumentText.Split('&', StringSplitOptions.RemoveEmptyEntries))
|
||||
{
|
||||
var separatorIndex = pair.IndexOf('=');
|
||||
if (separatorIndex < 0)
|
||||
{
|
||||
arguments[Uri.UnescapeDataString(pair)] = "";
|
||||
continue;
|
||||
}
|
||||
|
||||
var key = Uri.UnescapeDataString(pair[..separatorIndex]);
|
||||
var value = Uri.UnescapeDataString(pair[(separatorIndex + 1)..]);
|
||||
arguments[key] = value;
|
||||
}
|
||||
|
||||
return arguments;
|
||||
}
|
||||
|
||||
private static bool TryGetArgument(
|
||||
IReadOnlyDictionary<string, string> arguments,
|
||||
string key,
|
||||
out string value)
|
||||
{
|
||||
if (arguments.TryGetValue(key, out value!))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
value = "";
|
||||
return false;
|
||||
}
|
||||
|
||||
private static string FormatDuration(TimeSpan duration)
|
||||
{
|
||||
if (duration.TotalMinutes >= 1)
|
||||
{
|
||||
return $"{Math.Round(duration.TotalMinutes):0} minutes";
|
||||
}
|
||||
|
||||
return $"{Math.Max(1, Math.Round(duration.TotalSeconds)):0} seconds";
|
||||
}
|
||||
|
||||
private sealed record PendingPrompt(
|
||||
Func<MeetingInactivityPromptResponse, CancellationToken, Task> HandleResponseAsync);
|
||||
}
|
||||
#endif
|
||||
Reference in New Issue
Block a user