ci: add Claude Code Action workflow

[Alex-Wengg]

Summary

Adds .github/workflows/claude.yml so the repo can respond to @claude mentions in issues, issue comments, PR reviews, and PR review comments via anthropics/claude-code-action@v1.

Motivation: PR #596 had a reviewer post @claude review and nothing happened because no workflow was wired up. This PR fixes that for future reviews.

What it does

• Triggers on issue_comment, pull_request_review_comment, pull_request_review, issues (opened/assigned)
• Job runs only when the body/title contains @claude (cheap filter, prevents wasted runs)
• Uses ANTHROPIC_API_KEY repo secret for auth
• Minimal read permissions on contents/PRs/issues; id-token: write for OIDC

Required configuration (repo settings)

Before this workflow can run, a maintainer needs to:

1. Install the Claude GitHub App on FluidInference/FluidAudio
2. Add an ANTHROPIC_API_KEY secret in repo Settings -> Secrets and variables -> Actions

Without those, the workflow file is inert (no failed runs, just no-op).

Test plan

• Maintainer installs the Claude GitHub App and sets ANTHROPIC_API_KEY
• After merge, post @claude help on a throwaway issue and confirm the workflow fires
• Confirm non-@claude comments do not trigger the job

🤖 Generated with Claude Code

[github-actions]

Offline VBx Pipeline Results

Speaker Diarization Performance (VBx Batch Mode)

Optimal clustering with Hungarian algorithm for maximum accuracy

Metric	Value	Target	Status	Description
DER	10.4%	<20%	✅	Diarization Error Rate (lower is better)
RTFx	10.22x	>1.0x	✅	Real-Time Factor (higher is faster)

Offline VBx Pipeline Timing Breakdown

Time spent in each stage of batch diarization

Stage	Time (s)	%	Description
Model Download	11.897	11.6	Fetching diarization models
Model Compile	5.099	5.0	CoreML compilation
Audio Load	0.056	0.1	Loading audio file
Segmentation	25.042	24.4	VAD + speech detection
Embedding	102.333	99.6	Speaker embedding extraction
Clustering (VBx)	0.148	0.1	Hungarian algorithm + VBx clustering
Total	102.700	100	Full VBx pipeline

Speaker Diarization Research Comparison

Offline VBx achieves competitive accuracy with batch processing

Method	DER	Mode	Description
FluidAudio (Offline)	10.4%	VBx Batch	On-device CoreML with optimal clustering
FluidAudio (Streaming)	17.7%	Chunk-based	First-occurrence speaker mapping
Research baseline	18-30%	Various	Standard dataset performance

Pipeline Details:

• Mode: Offline VBx with Hungarian algorithm for optimal speaker-to-cluster assignment
• Segmentation: VAD-based voice activity detection
• Embeddings: WeSpeaker-compatible speaker embeddings
• Clustering: PowerSet with VBx refinement
• Accuracy: Higher than streaming due to optimal post-hoc mapping

_{🎯 Offline VBx Test • AMI Corpus ES2004a • 1049.0s meeting audio • 127.5s processing • Test runtime: 2m 10s • 05/11/2026, 11:14 AM EST}

[github-actions]

ASR Benchmark Results ✅

Status: All benchmarks passed

Parakeet v3 (multilingual)

Dataset	WER Avg	WER Med	RTFx	Status
test-clean	0.57%	0.00%	4.25x	✅
test-other	1.35%	0.00%	3.36x	✅

Parakeet v2 (English-optimized)

Dataset	WER Avg	WER Med	RTFx	Status
test-clean	0.80%	0.00%	3.69x	✅
test-other	1.00%	0.00%	2.50x	✅

Streaming (v3)

Metric	Value	Description
WER	0.00%	Word Error Rate in streaming mode
RTFx	0.50x	Streaming real-time factor
Avg Chunk Time	1.810s	Average time to process each chunk
Max Chunk Time	2.764s	Maximum chunk processing time
First Token	2.091s	Latency to first transcription token
Total Chunks	31	Number of chunks processed

Streaming (v2)

Metric	Value	Description
WER	0.00%	Word Error Rate in streaming mode
RTFx	0.47x	Streaming real-time factor
Avg Chunk Time	1.890s	Average time to process each chunk
Max Chunk Time	2.823s	Maximum chunk processing time
First Token	1.823s	Latency to first transcription token
Total Chunks	31	Number of chunks processed

_{Streaming tests use 5 files with 0.5s chunks to simulate real-time audio streaming}

_{25 files per dataset • Test runtime: 8m31s • 05/11/2026, 11:20 AM EST}

_{RTFx = Real-Time Factor (higher is better) • Calculated as: Total audio duration ÷ Total processing time
Processing time includes: Model inference on Apple Neural Engine, audio preprocessing, state resets between files, token-to-text conversion, and file I/O
Example: RTFx of 2.0x means 10 seconds of audio processed in 5 seconds (2x faster than real-time)}

Expected RTFx Performance on Physical M1 Hardware:

• M1 Mac: ~28x (clean), ~25x (other)
• CI shows ~0.5-3x due to virtualization limitations

_{Testing methodology follows HuggingFace Open ASR Leaderboard}

[github-actions]

PocketTTS Smoke Test ✅

Check	Result
Build	✅
Model download	✅
Model load	✅
Synthesis pipeline	✅
Output WAV	✅ (146.3 KB)

_{Runtime: 0m28s}

_{Note: PocketTTS uses CoreML MLState (macOS 15) KV cache + Mimi streaming state. CI VM lacks physical GPU — audio quality and performance may differ from Apple Silicon.}

[github-actions]

VAD Benchmark Results

Performance Comparison

Dataset	Accuracy	Precision	Recall	F1-Score	RTFx	Files
MUSAN	92.0%	86.2%	100.0%	92.6%	674.7x faster	50
VOiCES	92.0%	86.2%	100.0%	92.6%	542.4x faster	50

Dataset Details

• MUSAN: Music, Speech, and Noise dataset - standard VAD evaluation
• VOiCES: Voices Obscured in Complex Environmental Settings - tests robustness in real-world conditions

✅: Average F1-Score above 70%

[github-actions]

Sortformer High-Latency Benchmark Results

ES2004a Performance (30.4s latency config)

Metric	Value	Target	Status
DER	30.3%	<35%	✅
Miss Rate	28.2%	-	-
False Alarm	0.9%	-	-
Speaker Error	1.2%	-	-
RTFx	9.1x	>1.0x	✅
Speakers	4/4	-	-

_{Sortformer High-Latency • ES2004a • Runtime: 3m 41s • 2026-05-11T15:34:18.369Z}

[github-actions]

Kokoro TTS Smoke Test ✅

Check	Result
Build	✅
Model download	✅
Model load	✅
Synthesis pipeline	✅
Output WAV	✅ (634.8 KB)

_{Runtime: 0m51s}

_{Note: Kokoro TTS uses CoreML flow matching + Vocos vocoder. CI VM lacks physical ANE — performance may differ from Apple Silicon.}

[github-actions]

Qwen3-ASR int8 Smoke Test ✅

Check	Result
Build	✅
Model download	✅
Model load	✅
Transcription pipeline	✅
Decoder size	571 MB (vs 1.1 GB f32)

Performance Metrics

Metric	CI Value	Expected on Apple Silicon
Median RTFx	0.04x	~2.5x
Overall RTFx	0.04x	~2.5x

_{Runtime: 5m10s}

_{Note: CI VM lacks physical GPU — CoreML MLState (macOS 15) KV cache produces degraded results on virtualized runners. On Apple Silicon: ~1.3% WER / 2.5x RTFx.}

[github-actions]

Speaker Diarization Benchmark Results

Speaker Diarization Performance

Evaluating "who spoke when" detection accuracy

Metric	Value	Target	Status	Description
DER	15.1%	<30%	✅	Diarization Error Rate (lower is better)
JER	24.9%	<25%	✅	Jaccard Error Rate
RTFx	23.69x	>1.0x	✅	Real-Time Factor (higher is faster)

Diarization Pipeline Timing Breakdown

Time spent in each stage of speaker diarization

Stage	Time (s)	%	Description
Model Download	8.641	19.5	Fetching diarization models
Model Compile	3.703	8.4	CoreML compilation
Audio Load	0.072	0.2	Loading audio file
Segmentation	13.276	30.0	Detecting speech regions
Embedding	22.127	50.0	Extracting speaker voices
Clustering	8.851	20.0	Grouping same speakers
Total	44.286	100	Full pipeline

Speaker Diarization Research Comparison

Research baselines typically achieve 18-30% DER on standard datasets

Method	DER	Notes
FluidAudio	15.1%	On-device CoreML
Research baseline	18-30%	Standard dataset performance

Note: RTFx shown above is from GitHub Actions runner. On Apple Silicon with ANE:

• M2 MacBook Air (2022): Runs at 150 RTFx real-time
• Performance scales with Apple Neural Engine capabilities

_{🎯 Speaker Diarization Test • AMI Corpus ES2004a • 1049.0s meeting audio • 44.3s diarization time • Test runtime: 3m 17s • 05/11/2026, 11:49 AM EST}

[github-actions]

Parakeet EOU Benchmark Results ✅

Status: Benchmark passed
Chunk Size: 320ms
Files Tested: 100/100

Performance Metrics

Metric	Value	Description
WER (Avg)	7.03%	Average Word Error Rate
WER (Med)	4.17%	Median Word Error Rate
RTFx	7.57x	Real-time factor (higher = faster)
Total Audio	470.6s	Total audio duration processed
Total Time	66.5s	Total processing time

Streaming Metrics

Metric	Value	Description
Avg Chunk Time	0.066s	Average chunk processing time
Max Chunk Time	0.133s	Maximum chunk processing time
EOU Detections	0	Total End-of-Utterance detections

_{Test runtime: 1m25s • 05/11/2026, 11:51 AM EST}

_{RTFx = Real-Time Factor (higher is better) • Processing includes: Model inference, audio preprocessing, state management, and file I/O}