fix(flow): improve tail capture and final chunk ASR recovery
- Add FlowUtteranceEndCoordinator with 350ms silence drain and 150ms post-roll - Extend FinalChunkRecovery for short/empty final chunks in chunked pipeline - Snapshot partial at mic stop and guard final transcript in FlowSessionManager - Unify tail drain presets (iosFlow/macMLX) and expand diagnostics Co-authored-by: Rocky <hkgood@users.noreply.github.com>
This commit is contained in:
@@ -101,6 +101,7 @@ public actor ChunkedUtterancePipeline {
|
||||
var processedChunks = 0
|
||||
var previousChunkSamples: [Float] = []
|
||||
var lastChunkSamples = 0
|
||||
var didRetryEmptyFinal = false
|
||||
|
||||
let feeder = Task {
|
||||
for await chunk in UtteranceStreamChunker.chunks(from: stream, config: config) {
|
||||
@@ -118,20 +119,28 @@ public actor ChunkedUtterancePipeline {
|
||||
|
||||
guard let chunk = await queue.dequeue() else { break }
|
||||
|
||||
if chunk.isLast && chunk.samples.isEmpty {
|
||||
continue
|
||||
}
|
||||
|
||||
processedChunks += 1
|
||||
lastChunkSamples = chunk.samples.count
|
||||
|
||||
if chunk.isLast,
|
||||
chunk.samples.count < config.minFinalChunkSamples,
|
||||
processedChunks > 1,
|
||||
!previousChunkSamples.isEmpty {
|
||||
let mergedSamples = Array(previousChunkSamples.suffix(config.overlapSamples))
|
||||
+ chunk.samples
|
||||
let mergedResult = await transcribeChunk(samples: mergedSamples)
|
||||
if let preMerge = FinalChunkRecovery.preMergePlan(
|
||||
chunk: chunk,
|
||||
processedChunks: processedChunks,
|
||||
previousChunkSamples: previousChunkSamples,
|
||||
config: config
|
||||
) {
|
||||
FlowPipelineDiagnostics.logFinalChunkRecovery(
|
||||
action: "preMerge",
|
||||
chunkIndex: chunk.index
|
||||
)
|
||||
let mergedResult = await transcribeChunk(samples: preMerge.samples)
|
||||
switch mergedResult {
|
||||
case .success(let text):
|
||||
stitcher.removeLastSegment()
|
||||
stitcher.append(index: max(0, chunk.index - 1), text: text)
|
||||
stitcher.append(index: preMerge.stitchIndex, text: text)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
case .failure(let message):
|
||||
failedChunks += 1
|
||||
@@ -153,8 +162,52 @@ public actor ChunkedUtterancePipeline {
|
||||
let result = await transcribeChunk(samples: chunk.samples)
|
||||
switch result {
|
||||
case .success(let text):
|
||||
stitcher.append(index: chunk.index, text: text)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
if chunk.isLast,
|
||||
!didRetryEmptyFinal,
|
||||
let retry = FinalChunkRecovery.emptyResultRetryPlan(
|
||||
chunk: chunk,
|
||||
previousChunkSamples: previousChunkSamples,
|
||||
config: config,
|
||||
asrText: text
|
||||
) {
|
||||
didRetryEmptyFinal = true
|
||||
FlowPipelineDiagnostics.logFinalChunkRecovery(
|
||||
action: "emptyRetry",
|
||||
chunkIndex: chunk.index
|
||||
)
|
||||
let retryResult = await transcribeChunk(samples: retry.samples)
|
||||
switch retryResult {
|
||||
case .success(let retryText):
|
||||
let trimmed = retryText.trimmingCharacters(in: .whitespacesAndNewlines)
|
||||
if !trimmed.isEmpty {
|
||||
if retry.stitchIndex < chunk.index {
|
||||
stitcher.removeLastSegment()
|
||||
}
|
||||
stitcher.append(index: retry.stitchIndex, text: retryText)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
} else {
|
||||
stitcher.append(index: chunk.index, text: text)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
}
|
||||
case .failure(let message):
|
||||
stitcher.append(index: chunk.index, text: text)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
failedChunks += 1
|
||||
chunkWarnings.append(
|
||||
SharedL10n.format(
|
||||
"error.asr.chunkFailed",
|
||||
chunk.index + 1,
|
||||
message
|
||||
)
|
||||
)
|
||||
case .cancelled:
|
||||
feeder.cancel()
|
||||
return .cancelled
|
||||
}
|
||||
} else {
|
||||
stitcher.append(index: chunk.index, text: text)
|
||||
publishPartial(from: stitcher, onPartial: onPartial)
|
||||
}
|
||||
case .failure(let message):
|
||||
failedChunks += 1
|
||||
chunkWarnings.append(
|
||||
|
||||
@@ -666,16 +666,11 @@ public final class FlowContinuousCapture {
|
||||
gate.withLock { $0 = .draining }
|
||||
drainTracker.beginDrain()
|
||||
|
||||
var endedBySilence = false
|
||||
while true {
|
||||
let decision = drainTracker.shouldFinish(policy: policy)
|
||||
if decision.finished {
|
||||
endedBySilence = decision.endedBySilence
|
||||
break
|
||||
}
|
||||
if Task.isCancelled { break }
|
||||
try? await Task.sleep(nanoseconds: FlowCaptureConstants.drainPollIntervalNs)
|
||||
}
|
||||
let timing = await FlowUtteranceEndCoordinator.awaitTailCapture(
|
||||
tracker: drainTracker,
|
||||
policy: policy,
|
||||
pollIntervalNs: FlowCaptureConstants.drainPollIntervalNs
|
||||
)
|
||||
|
||||
// NOTE: We intentionally do NOT signal `.endOfStream` to the shared
|
||||
// downsampling converter here. `AVAudioConverter` is stateful: once its
|
||||
@@ -692,8 +687,9 @@ public final class FlowContinuousCapture {
|
||||
let tailSamples = tailSampleCounter.withLock { $0 }
|
||||
let report = FlowCaptureDrainReport(
|
||||
drainDurationSeconds: drainTracker.elapsedSeconds(),
|
||||
endedBySilence: endedBySilence,
|
||||
tailSampleCount: tailSamples
|
||||
endedBySilence: timing.endedBySilence,
|
||||
tailSampleCount: tailSamples,
|
||||
postRollDurationSeconds: timing.postRollDurationSeconds
|
||||
)
|
||||
drainTracker.reset()
|
||||
tailSampleCounter.withLock { $0 = 0 }
|
||||
|
||||
@@ -558,12 +558,10 @@ public final class LiveDictationController: ObservableObject {
|
||||
drainTracker.beginDrain()
|
||||
|
||||
let policy = FlowCaptureTailDrainPolicy.flowDefault
|
||||
while true {
|
||||
let decision = drainTracker.shouldFinish(policy: policy)
|
||||
if decision.finished { break }
|
||||
if Task.isCancelled { break }
|
||||
try? await Task.sleep(nanoseconds: 20_000_000)
|
||||
}
|
||||
_ = await FlowUtteranceEndCoordinator.awaitTailCapture(
|
||||
tracker: drainTracker,
|
||||
policy: policy
|
||||
)
|
||||
|
||||
// Trailing speech is preserved by the live `.draining` forwarding
|
||||
// loop above. We deliberately do NOT signal `.endOfStream` to the
|
||||
|
||||
Reference in New Issue
Block a user