2024 MacBook Pro, 48GB Ram, M4 Pro, Tahoe 26.0
https://huggingface.co/FluidInference/parakeet-tdt-0.6b-v3-coreml
swift run fluidaudiocli fleurs-benchmark --languages all --samples allLanguage | WER% | CER% | RTFx | Duration | Processed | Skipped
-----------------------------------------------------------------------------------------
Bulgarian (Bulgaria) | 12.8 | 4.1 | 195.2 | 3468.0s | 350 | -
Croatian (Croatia) | 14.0 | 4.3 | 204.9 | 3647.0s | 350 | -
Czech (Czechia) | 12.0 | 3.8 | 214.2 | 4247.4s | 350 | -
Danish (Denmark) | 20.2 | 7.4 | 214.4 | 10579.1s | 930 | -
Dutch (Netherlands) | 7.8 | 2.6 | 191.7 | 3337.7s | 350 | -
English (US) | 5.4 | 2.5 | 207.4 | 3442.9s | 350 | -
Estonian (Estonia) | 20.1 | 4.2 | 225.3 | 10825.4s | 893 | -
Finnish (Finland) | 14.8 | 3.1 | 222.0 | 11894.4s | 918 | -
French (France) | 5.9 | 2.2 | 199.9 | 3667.3s | 350 | -
German (Germany) | 5.9 | 1.9 | 220.9 | 4684.6s | 350 | -
Greek (Greece) | 36.9 | 13.7 | 183.0 | 6862.0s | 650 | -
Hungarian (Hungary) | 17.6 | 5.2 | 213.6 | 11050.9s | 905 | -
Italian (Italy) | 4.0 | 1.3 | 236.7 | 5098.7s | 350 | -
Latvian (Latvia) | 27.1 | 7.5 | 217.8 | 10218.6s | 851 | -
Lithuanian (Lithuania) | 25.0 | 6.8 | 202.8 | 10686.5s | 986 | -
Maltese (Malta) | 25.2 | 9.3 | 217.4 | 12770.6s | 926 | -
Polish (Poland) | 8.6 | 2.8 | 190.2 | 3409.6s | 350 | -
Romanian (Romania) | 14.4 | 4.7 | 200.4 | 9099.4s | 883 | -
Russian (Russia) | 7.2 | 2.2 | 209.7 | 3974.6s | 350 | -
Slovak (Slovakia) | 12.6 | 4.4 | 227.6 | 4169.6s | 350 | -
Slovenian (Slovenia) | 27.4 | 9.2 | 197.1 | 8173.1s | 834 | -
Spanish (Spain) | 4.5 | 2.2 | 221.7 | 4258.9s | 350 | -
Swedish (Sweden) | 16.8 | 5.0 | 219.5 | 8399.2s | 759 | -
Ukrainian (Ukraine) | 7.2 | 2.5 | 201.9 | 3853.7s | 350 | -
-----------------------------------------------------------------------------------------
AVERAGE | 14.7 | 4.7 | 209.8 | 161819.2 | 14085 | -
Dataset: librispeech test-clean
Files processed: 2620
Average WER: 2.5%
Median WER: 0.0%
Average CER: 1.0%
Median RTFx: 139.6x
Overall RTFx: 155.6x (19452.5s / 125.0s)
swift run fluidaudiocli asr-benchmark --max-files all --model-version v2
Use v2 if you only need English, it is a bit more accurate
--- Benchmark Results ---
Dataset: librispeech test-clean
Files processed: 2620
Average WER: 2.1%
Median WER: 0.0%
Average CER: 0.7%
Median RTFx: 128.6x
Overall RTFx: 145.8x (19452.5s / 133.4s)
Core ML first-load compile times captured on iPhone 16 Pro Max and iPhone 13 running the parakeet-tdt-0.6b-v3-coreml bundle. Cold-start compilation happens the first time each Core ML model is loaded; subsequent loads hit the cached binaries. Warm compile metrics were collected only on the iPhone 16 Pro Max run, and only for models that were reloaded during the session.
| Model | iPhone 16 Pro Max cold (ms) | iPhone 16 Pro Max warm (ms) | iPhone 13 cold (ms) | Compute units |
|---|---|---|---|---|
| Preprocessor | 9.15 | - | 632.63 | MLComputeUnits(rawValue: 2) |
| Encoder | 3361.23 | 162.05 | 4396.00 | MLComputeUnits(rawValue: 1) |
| Decoder | 88.49 | 8.11 | 146.01 | MLComputeUnits(rawValue: 1) |
| JointDecision | 48.46 | 7.97 | 71.85 | MLComputeUnits(rawValue: 1) |
CTC-based custom vocabulary boosting system, which enables accurate recognition of domain-specific terms (company names, technical jargon, proper nouns) without retraining the ASR model.
# Download the dataset
swift run fluidaudiocli ctc-earnings-benchmark --auto-download
# Run the benchmark
swift run fluidaudiocli ctc-earnings-benchmark
Earnings Benchmark (TDT transcription + CTC keyword spotting)
Data directory: /Users/<user>/Library/Application Support/FluidAudio/earnings22-kws/test-dataset
Output file: ctc_earnings_benchmark.json
TDT version: v2
CTC model: /Users/<user>/Library/Application Support/FluidAudio/Models/parakeet-ctc-110m-coreml
Loading TDT models (v2) for transcription...
TDT models loaded successfully
Loading CTC models from: /Users/<user>/Library/Application Support/FluidAudio/Models/parakeet-ctc-110m-coreml
Loaded CTC vocabulary with 1024 tokens, variant: Parakeet CTC 110M (hybrid)
Created CTC spotter with blankId=1024
Processing 773 test files...
[ 1/772] 4329526_chunk0 WER: 10.3% Dict: 1/1
[ 2/772] 4329526_chunk109 WER: 12.5% Dict: 2/2
[ 3/772] 4329526_chunk118 WER: 3.1% Dict: 3/3
[ 4/772] 4329526_chunk132 WER: 8.1% Dict: 1/1
[ 5/772] 4329526_chunk135 WER: 25.7% Dict: 1/1
[ 6/772] 4329526_chunk16 WER: 8.6% Dict: 1/1
...
[767/772] 4485206_chunk_86 WER: 5.0% Dict: 2/2
[768/772] 4485206_chunk_88 WER: 8.3% Dict: 2/2
[769/772] 4485206_chunk_92 WER: 14.7% Dict: 4/4
[770/772] 4485206_chunk_97 WER: 30.5% Dict: 1/1
[771/772] 4485206_chunk_98 WER: 18.6% Dict: 4/4
[772/772] 4485206_chunk_99 WER: 22.0% Dict: 1/1
============================================================
EARNINGS22 BENCHMARK (TDT + CTC)
============================================================
Model: /Users/<user>/Library/Application Support/FluidAudio/Models/parakeet-ctc-110m-coreml
Total tests: 771
Average WER: 15.00%
Dict Pass (Recall): 1299/1308 (99.3%)
Vocab Precision: 99.3% (TP=1068, FP=8)
Vocab Recall: 85.2% (TP=1068, FN=185)
Vocab F-score: 91.7%
Total audio: 11564.5s
Total processing: 182.5s
RTFx: 63.36x
============================================================
Results written to: ctc_earnings_benchmark.jsonIn context of vocabulary/keyword detection:
| Metric | Definition |
|---|---|
| TP (True Positive) | Word is in reference AND in hypothesis (correctly detected) |
| FP (False Positive) | Word is in hypothesis but NOT in reference (hallucinated/wrong) |
| FN (False Negative) | Word is in reference but NOT in hypothesis (missed) |
Derived metrics:
| Metric | Formula | Meaning |
|---|---|---|
| Precision | TP / (TP + FP) | "Of words we output, how many were correct?" |
| Recall | TP / (TP + FN) | "Of words that should appear, how many did we find?" |
| F-Score | 2 × P × R / (P + R) | Harmonic mean of precision and recall |
We generated the same strings with to generate audio between 1s to ~300s in order to test the speed across a range of varying inputs on Pytorch CPU, MPS, and MLX pipeline, and compared it against the native Swift version with Core ML models.
Each pipeline warmed up the models by running through it once with pesudo inputs, and then comparing the raw inference time with the model already loaded. You can see that for the Core ML model, we traded lower memory and very slightly faster inference for longer initial warm-up.
Note that the Pytorch kokoro model in Pytorch has a memory leak issue: hexgrad/kokoro#152
The following tests were ran on M4 Pro, 48GB RAM, Macbook Pro. If you have another device, please do try replicating it as well!
KPipeline benchmark for voice af_heart (warm-up took 0.175s) using hexgrad/kokoro
Test Chars Output (s) Inf(s) RTFx Peak GB
1 42 2.750 0.187 14.737x 1.44
2 129 8.625 0.530 16.264x 1.85
3 254 15.525 0.923 16.814x 2.65
4 93 6.125 0.349 17.566x 2.66
5 104 7.200 0.410 17.567x 2.70
6 130 9.300 0.504 18.443x 2.72
7 197 12.850 0.726 17.711x 2.83
8 6 1.350 0.098 13.823x 2.83
9 1228 76.200 4.342 17.551x 3.19
10 567 35.200 2.069 17.014x 4.85
11 4615 286.525 17.041 16.814x 4.78
Total - 461.650 27.177 16.987x 4.85 I wasn't able to run the MPS model for longer durations, even with PYTORCH_ENABLE_MPS_FALLBACK=1 enabled, it kept crashing for the longer strings.
KPipeline benchmark for voice af_heart (warm-up took 0.568s) using pip package
Test Chars Output (s) Inf(s) RTFx Peak GB
1 42 2.750 0.414 6.649x 1.41
2 129 8.625 0.729 11.839x 1.54
Total - 11.375 1.142 9.960x 1.54 TTS benchmark for voice af_heart (warm-up took an extra 2.155s) using model prince-canuma/Kokoro-82M
Test Chars Output (s) Inf(s) RTFx Peak GB
1 42 2.750 0.347 7.932x 1.12
2 129 8.650 0.597 14.497x 2.47
3 254 15.525 0.825 18.829x 2.65
4 93 6.125 0.306 20.039x 2.65
5 104 7.200 0.343 21.001x 2.65
6 130 9.300 0.560 16.611x 2.65
7 197 12.850 0.596 21.573x 2.65
8 6 1.350 0.364 3.706x 2.65
9 1228 76.200 2.979 25.583x 3.29
10 567 35.200 1.374 25.615x 3.37
11 4615 286.500 11.112 25.783x 3.37
Total - 461.650 19.401 23.796x 3.37Note that it does take ~15s to compile the model on the first run, subsequent runs are shorter, we expect ~2s to load.
> swift run fluidaudiocli tts --benchmark
...
FluidAudio TTS benchmark for voice af_heart (warm-up took an extra 2.348s)
Test Chars Ouput (s) Inf(s) RTFx
1 42 2.825 0.440 6.424x
2 129 7.725 0.594 13.014x
3 254 13.400 0.776 17.278x
4 93 5.875 0.587 10.005x
5 104 6.675 0.613 10.889x
6 130 8.075 0.621 13.008x
7 197 10.650 0.627 16.983x
8 6 0.825 0.360 2.290x
9 1228 67.625 2.362 28.625x
10 567 33.025 1.341 24.619x
11 4269 247.600 9.087 27.248x
Total - 404.300 17.408 23.225
Peak memory usage (process-wide): 1.503 GBModel is nearly identical to the base model in terms of quality, performance wise we see an up to ~3.5x improvement compared to the silero Pytorch VAD model with the 256ms batch model (8 chunks of 32ms)
Dataset: https://github.com/Lab41/VOiCES-subset
swift run fluidaudiocli vad-benchmark --dataset voices-subset --all-files --threshold 0.85
...
Timing Statistics:
[18:56:31.208] [INFO] [VAD] Total processing time: 0.29s
[18:56:31.208] [INFO] [VAD] Total audio duration: 351.05s
[18:56:31.208] [INFO] [VAD] RTFx: 1230.6x faster than real-time
[18:56:31.208] [INFO] [VAD] Audio loading time: 0.00s (0.6%)
[18:56:31.208] [INFO] [VAD] VAD inference time: 0.28s (98.7%)
[18:56:31.208] [INFO] [VAD] Average per file: 0.011s
[18:56:31.208] [INFO] [VAD] Min per file: 0.001s
[18:56:31.208] [INFO] [VAD] Max per file: 0.020s
[18:56:31.208] [INFO] [VAD]
VAD Benchmark Results:
[18:56:31.208] [INFO] [VAD] Accuracy: 96.0%
[18:56:31.208] [INFO] [VAD] Precision: 100.0%
[18:56:31.208] [INFO] [VAD] Recall: 95.8%
[18:56:31.208] [INFO] [VAD] F1-Score: 97.9%
[18:56:31.208] [INFO] [VAD] Total Time: 0.29s
[18:56:31.208] [INFO] [VAD] RTFx: 1230.6x faster than real-time
[18:56:31.208] [INFO] [VAD] Files Processed: 25
[18:56:31.208] [INFO] [VAD] Avg Time per File: 0.011s
swift run fluidaudiocli vad-benchmark --dataset musan-full --num-files all --threshold 0.8
...
[23:02:35.539] [INFO] [VAD] Total processing time: 322.31s
[23:02:35.539] [INFO] [VAD] Timing Statistics:
[23:02:35.539] [INFO] [VAD] RTFx: 1220.7x faster than real-time
[23:02:35.539] [INFO] [VAD] Audio loading time: 1.20s (0.4%)
[23:02:35.539] [INFO] [VAD] VAD inference time: 319.57s (99.1%)
[23:02:35.539] [INFO] [VAD] Average per file: 0.160s
[23:02:35.539] [INFO] [VAD] Total audio duration: 393442.58s
[23:02:35.539] [INFO] [VAD] Min per file: 0.000s
[23:02:35.539] [INFO] [VAD] Max per file: 0.873s
[23:02:35.711] [INFO] [VAD] VAD Benchmark Results:
[23:02:35.711] [INFO] [VAD] Accuracy: 94.2%
[23:02:35.711] [INFO] [VAD] Precision: 92.6%
[23:02:35.711] [INFO] [VAD] Recall: 78.9%
[23:02:35.711] [INFO] [VAD] F1-Score: 85.2%
[23:02:35.711] [INFO] [VAD] Total Time: 322.31s
[23:02:35.711] [INFO] [VAD] RTFx: 1220.7x faster than real-time
[23:02:35.711] [INFO] [VAD] Files Processed: 2016
[23:02:35.711] [INFO] [VAD] Avg Time per File: 0.160s
[23:02:35.744] [INFO] [VAD] Results saved to: vad_benchmark_results.json
Encoder-decoder ASR using Qwen3-ASR-0.6B converted to CoreML. Autoregressive generation with KV-cache.
Note: WER/CER may be higher than the original PyTorch model due to CoreML conversion limitations. See FLEURS results below for full multilingual benchmarks.
Model: FluidInference/qwen3-asr-0.6b-coreml (f32 variant)
Hardware: Apple M2, 2022, macOS 26
| Metric | Value |
|---|---|
| WER (Avg) | 4.4% |
| WER (Median) | 0.0% |
| RTFx | 2.8x |
| Per-token | ~75ms |
| Metric | Value |
|---|---|
| CER (Avg) | 6.6% |
| WER (Avg) | 10.3% |
| Median RTFx | 4.6x |
| Overall RTFx | 3.8x |
| Processing Time | 2.6h |
Methodology notes:
- CER (Character Error Rate) is the primary metric for Chinese ASR, as per the Qwen3-ASR Technical Report: "We use CER for character-based languages (e.g., Mandarin Chinese, Cantonese, and Korean) and WER for word-delimited languages"
- WER calculation uses Apple's
NLTokenizerfor Chinese word segmentation; we were unable to verify how official Qwen3-ASR evaluation performs tokenization - Official Qwen3-ASR reports 3.15% on AISHELL-2 (different dataset) per HuggingFace model card; our 6.6% CER on AISHELL-1 suggests some accuracy loss in CoreML conversion
- Why AISHELL-1? AISHELL-2 (1000h) requires an application with institutional affiliation and is restricted to non-commercial use. AISHELL-1 (178h) is openly available under Apache 2.0.
- Dataset: AudioLLMs/aishell_1_zh_test
# Run AISHELL-1 benchmark
swift run -c release fluidaudiocli qwen3-benchmark --dataset aishellFull benchmark across all 30 languages supported by Qwen3-ASR, matching the official FLEURS tiers.
Which metric to use:
- CER for character-based languages (Chinese, Japanese, Korean, Thai, Vietnamese, Cantonese) - WER is meaningless due to word segmentation differences
- WER for word-delimited languages (European, Arabic, etc.)
| Tier | Languages | Our CER | Official 0.6B WER |
|---|---|---|---|
| FLEURS (12 core) | en, zh, yue, ar, de, es, fr, it, ja, ko, pt, ru | 10.3% | 10.0% |
| FLEURS† (8 add) | hi, id, ms, nl, pl, th, tr, vi | 20.9% | 31.9% |
| FLEURS†† (10 hardest) | cs, da, el, fa, fi, fil, hu, mk, ro, sv | 41.0% | 47.8% |
Note: Official Qwen3-ASR reports WER, but for CJK languages this includes word segmentation artifacts. Our CER comparison shows CoreML conversion has minimal accuracy loss on core languages.
| Language | RTFx | Avg CER | Med CER | Avg WER | Med WER | Use |
|---|---|---|---|---|---|---|
| en_us | 1.16x | 4.0% | 2.3% | 7.3% | 5.3% | WER |
| es_419 | 2.04x | 4.9% | 3.0% | 10.5% | 8.1% | WER |
| it_it | 3.46x | 5.1% | 2.8% | 12.4% | 10.0% | WER |
| ru_ru | 1.84x | 6.9% | 4.6% | 18.0% | 15.6% | WER |
| de_de | 1.22x | 8.1% | 5.1% | 16.6% | 13.3% | WER |
| pt_br | 3.27x | 8.6% | 5.4% | 17.5% | 13.0% | WER |
| fr_fr | 1.72x | 8.9% | 6.2% | 17.3% | 13.3% | WER |
| cmn_hans_cn | 1.74x | 9.4% | 5.1% | 99.7%* | 100%* | CER |
| ko_kr | 1.10x | 10.6% | 7.9% | 23.5% | 21.7% | CER |
| tr_tr | 2.84x | 11.6% | 9.6% | 33.0% | 31.2% | WER |
| id_id | 2.86x | 16.0% | 9.1% | 30.9% | 22.2% | WER |
| nl_nl | 2.29x | 17.2% | 13.6% | 36.5% | 30.3% | WER |
| ms_my | 2.24x | 17.4% | 13.2% | 37.6% | 33.3% | WER |
| th_th | 1.42x | 18.3% | 15.4% | 96.8%* | 100%* | CER |
| ar_eg | 1.53x | 18.5% | 13.8% | 40.3% | 36.4% | WER |
| ja_jp | 0.83x | 19.3% | 17.1% | 94.4%* | 100%* | CER |
| yue_hant_hk | 0.87x | 19.5% | 13.8% | 99.8%* | 100%* | CER |
| vi_vn | 2.69x | 25.4% | 21.0% | 35.9% | 31.0% | CER |
| fi_fi | 1.56x | 25.9% | 22.7% | 70.3% | 70.0% | WER |
| hi_in | 0.74x | 30.8% | 21.4% | 36.0% | 30.6% | WER |
| pl_pl | 1.69x | 30.8% | 27.4% | 61.9% | 60.0% | WER |
| sv_se | 2.38x | 31.3% | 30.1% | 67.8% | 66.7% | WER |
| fil_ph | 1.56x | 32.2% | 22.4% | 64.8% | 61.1% | WER |
| mk_mk | 0.79x | 43.2% | 27.9% | 73.0% | 75.9% | WER |
| da_dk | 2.33x | 45.5% | 46.5% | 81.1% | 84.6% | WER |
| fa_ir | 1.88x | 48.9% | 34.4% | 75.1% | 75.0% | WER |
| el_gr | 0.95x | 51.9% | 39.2% | 78.2% | 76.5% | WER |
| hu_hu | 1.05x | 59.0% | 55.7% | 91.8% | 95.8% | WER |
| ro_ro | 1.03x | 60.9% | 56.2% | 97.2% | 100% | WER |
| cs_cz | 2.26x | 62.2% | 56.5% | 88.2% | 96.2% | WER |
*WER >90% is expected for CJK/Thai due to word segmentation - FLEURS references have artificial character-by-character spacing while our output is natural continuous text. CER shows actual transcription quality.
| Metric | Average | Median |
|---|---|---|
| CER (all 30) | 25.1% | 19.4% |
| RTFx | 1.78x | 1.72x |
| Type | Avg RTFx | Notes |
|---|---|---|
| Romance (es, it, pt, fr) | 2.6x | Fastest |
| Turkic/Indonesian | 2.5x | Fast |
| Germanic (en, de, nl) | 1.6x | Medium |
| Slavic (ru, pl, cs) | 1.9x | Medium |
| CJK (zh, ja, ko, yue) | 1.1x | Slow - more tokens |
| Indic (hi) | 0.74x | Slowest |
# Run FLEURS benchmark for all languages
swift run -c release fluidaudiocli qwen3-benchmark --dataset fleurs --languages allReal-time streaming ASR with End-of-Utterance detection using the Parakeet EOU 120M CoreML model.
Model: FluidInference/parakeet-realtime-eou-120m-coreml
Hardware: Apple M2, 2022, macOS 26
| Chunk Size | WER (Avg) | Median WER | RTFx | Total Time |
|---|---|---|---|---|
| 320ms | 4.88% | 0.00% | 19.25x | 1015s (16.9m) |
| 160ms | 8.23% | 5.26% | 5.78x | 3387s (56.4m) |
# Run 320ms benchmark
swift run -c release fluidaudiocli parakeet-eou --benchmark --chunk-size 320 --use-cache
# Run 160ms benchmark
swift run -c release fluidaudiocli parakeet-eou --benchmark --chunk-size 160 --use-cacheNVIDIA's Nemotron Speech Streaming 0.6B model for low-latency streaming ASR.
Model: FluidInference/nemotron-speech-streaming-0.6b-coreml
Hardware: Apple M1, 2020, macOS 26
| Chunk Size | WER (Avg) | Median WER | RTFx | Total Time |
|---|---|---|---|---|
| 1120ms | 2.51% | 0.00% | 6.03x | 3228s (53.8m) |
| 560ms | 2.12% | 0.00% | TBD | TBD |
# Run 1120ms benchmark
swift run -c release fluidaudiocli nemotron-benchmark --chunk 1120
# Run 560ms benchmark
swift run -c release fluidaudiocli nemotron-benchmark --chunk 560The offline version uses the community-1 model, the online version uses the legacy speaker-diarization-3.1 model.
For slightly ~1.2% worse DER we default to a higher step ratio segmentation duration than the baseline community-1 pipeline. This allows us to get nearly ~2x the speed (as expected because we're processing 1/2 of the embeddings). For highly critical use cases, one may should use step ratio = 0.1 and minSegmentDurationSeconds = 0.0
Running on the full voxconverse benchmark:
StepRatio = 0.2, minSegmentDurationSeconds= 1.0
Average DER: 15.07% | Median DER: 10.70% | Average JER: 39.40% | Median JER: 40.95% (collar=0.25s, ignoreOverlap=True)
Average RTFx: 122.06 (from 232 clips)
Completed. New results: 232, Skipped existing: 0, Total attempted: 232
Step Ratio 2, min duration 1.0
StepRatio = 0.1, minSegmentDurationSeconds= 0
Average DER: 13.89% | Median DER: 10.49% | Average JER: 42.84% | Median JER: 43.30% (collar=0.25s, ignoreOverlap=True)
Average RTFx: 64.75 (from 232 clips)
Completed. New results: 232, Skipped existing: 0, Total attempted: 232
Step Ratio 1, min duration 0 (edited) Note that the baseline pytorch version is ~11% DER, we lost some precision dropping down to fp16 precision in order to run most of the embedding model on neural engine. But as a result, we significantly out perform the baseline mps backend as well. the pyannote-community-1 on cpu is ~1.5-2 RTFx, on mps, it's ~20-25 RTFx.
This is more tricky and honestly a lot more fragile to clustering. Expect +10-15% worse DER for the streaming implementation. Only use this when you critically need realtime streaming speaker diarization. In most cases, offline is more than enough for most applications.
Running a near real-time diarization benchmark for 3s chunks, 1s overlap, and 0.85 clustering threshold:
swift run fluidaudiocli diarization-benchmark --mode streaming \
--dataset ami-sdm \
--threshold 0.85 \
--auto-download \
--chunk-seconds 3.0 \
--overlap-seconds 1.0
...
------------------------------------------------------------------------------------------
Meeting DER % JER % Miss % FA % SE % Speakers RTFx
------------------------------------------------------------------------------------------
ES2004a 31.6 41.6 6.7 2.1 22.7 7/4 49.8
ES2005a 39.7 65.0 6.9 7.3 25.5 5/4 59.1
IS1002b 40.4 51.3 1.1 5.2 34.1 9/4 45.3
ES2002a 41.5 56.0 5.3 10.1 26.1 6/4 48.6
ES2003a 53.1 78.7 5.3 2.3 45.5 5/4 57.1
IS1000a 66.7 74.0 6.1 7.6 53.0 7/4 50.7
IS1001a 75.0 88.6 7.1 4.7 63.2 10/4 48.8
------------------------------------------------------------------------------------------
AVERAGE 49.7 65.0 5.5 5.6 38.6 - 51.4
==========================================================================================Diarization benchmark with 10s chunks, 0s overlap, and 0.7 clustering threshold:
swift run fluidaudiocli diarization-benchmark --mode streaming \
--dataset ami-sdm
--threshold 0.7
--auto-download
--chunk-seconds 10.0
--overlap-seconds 0.0
...
------------------------------------------------------------------------------------------
Meeting DER % JER % Miss % FA % SE % Speakers RTFx
------------------------------------------------------------------------------------------
ES2003a 12.0 19.5 6.9 1.2 3.9 4/4 477.0
ES2004a 15.1 24.8 9.2 1.2 4.7 4/4 367.4
ES2002a 17.8 26.8 8.6 5.8 3.4 6/4 356.8
IS1002b 38.0 41.8 3.1 3.1 31.8 5/4 361.9
ES2005a 22.5 36.8 7.7 6.8 8.0 4/4 460.8
IS1000a 57.7 80.6 11.9 3.9 41.9 8/4 352.1
IS1001a 70.1 85.4 11.2 2.4 56.5 7/4 370.9
------------------------------------------------------------------------------------------
AVERAGE 33.3 45.1 8.4 3.5 21.5 - 392.4
==========================================================================================Diarization benchmark with 5s chunks, 0s overlap, and 0.8 clustering threshold (best configuration found):
swift run fluidaudiocli diarization-benchmark --mode streaming \
--dataset ami-sdm
--threshold 0.8
--auto-download
--chunk-seconds 5.0
--overlap-seconds 0.0
...
------------------------------------------------------------------------------------------
Meeting DER % JER % Miss % FA % SE % Speakers RTFx
------------------------------------------------------------------------------------------
IS1002b 9.8 11.7 3.5 3.8 2.6 5/4 205.2
ES2003a 14.4 23.3 7.4 1.6 5.3 4/4 260.9
ES2004a 17.0 26.0 9.0 1.3 6.7 7/4 218.1
ES2005a 18.4 31.0 9.2 5.8 3.4 4/4 259.8
ES2002a 20.8 30.5 9.5 7.4 3.9 5/4 198.0
IS1000a 24.7 35.7 12.1 4.3 8.3 6/4 204.2
IS1001a 78.0 94.5 13.3 3.0 61.6 6/4 215.7
------------------------------------------------------------------------------------------
AVERAGE 26.2 36.1 9.2 3.9 13.1 - 223.1
==========================================================================================Diarization benchmark with 5s chunks, 2s overlap, and 0.8 clustering threshold:
swift run fluidaudiocli diarization-benchmark --mode streaming \
--dataset ami-sdm
--threshold 0.8
--auto-download
--chunk-seconds 5.0
--overlap-seconds 2.0
...
------------------------------------------------------------------------------------------
Meeting DER % JER % Miss % FA % SE % Speakers RTFx
------------------------------------------------------------------------------------------
ES2003a 24.5 42.1 4.7 1.9 18.0 6/4 81.4
ES2005a 27.5 50.6 5.5 7.6 14.4 5/4 76.8
ES2004a 31.6 54.8 6.4 2.3 23.0 5/4 66.9
IS1002b 39.6 57.0 0.8 5.1 33.7 6/4 63.7
ES2002a 41.1 57.2 4.7 9.8 26.7 5/4 65.5
IS1000a 57.4 54.2 6.1 7.7 43.6 9/4 67.2
IS1001a 79.0 86.8 7.0 5.0 66.9 10/4 64.5
------------------------------------------------------------------------------------------
AVERAGE 43.0 57.5 5.0 5.6 32.3 - 69.4
==========================================================================================NVIDIA's Sortformer model for streaming speaker diarization, converted to CoreML.
Model: FluidInference/diar-streaming-sortformer-coreml (V2 models for macOS 26+ compatibility)
Hardware: Apple M2, 2022, macOS 26.1
swift run fluidaudiocli sortformer-benchmark --nvidia-high-latency --hf --auto-download================================================================================
SORTFORMER BENCHMARK SUMMARY
================================================================================
Results Sorted by DER:
----------------------------------------------------------------------
Meeting DER % Miss % FA % SE % Speakers RTFx
----------------------------------------------------------------------
IS1009b 16.4 10.6 0.6 5.3 4/4 127.0
ES2004c 23.8 17.8 0.3 5.7 4/4 126.5
ES2004b 23.9 18.7 0.2 5.0 4/4 123.9
IS1009a 26.5 16.0 1.4 9.1 4/4 134.4
ES2004d 28.3 19.7 0.3 8.3 4/4 123.5
IS1009d 29.1 16.5 1.0 11.6 4/4 127.9
TS3003b 31.1 27.1 0.6 3.4 4/4 125.5
EN2002c 31.8 20.1 0.2 11.5 4/3 126.0
ES2004a 33.7 24.6 0.1 9.0 4/4 127.2
EN2002b 34.0 20.2 0.6 13.3 4/4 127.7
TS3003c 34.4 31.1 0.3 3.1 4/4 126.6
EN2002a 35.6 20.0 0.4 15.2 4/4 125.4
EN2002d 37.1 20.1 0.5 16.5 4/4 125.5
IS1009c 38.1 12.8 0.9 24.4 4/4 129.2
TS3003d 41.0 32.0 0.1 8.8 4/4 125.6
TS3003a 41.8 36.8 0.7 4.3 4/4 125.7
----------------------------------------------------------------------
AVERAGE 31.7 21.5 0.5 9.7 - 126.7
======================================================================
A research prototype from Westlake University for streaming speaker diarization.
Model: GradientDescent2718/ls-eend-coreml.
Hardware: Apple M4 MAX, 2026, macOS 26.1 (CPU only)
swift run fluidaudiocli lseend-benchmark --variant ami --auto-download================================================================================
LS-EEND BENCHMARK SUMMARY
[23:17:34.535] [DEBUG] [FluidAudio.LSEENDDiarizer] LS-EEND state reset
================================================================================
Results Sorted by DER:
----------------------------------------------------------------------
Meeting DER % Miss % FA % SE % Speakers RTFx
----------------------------------------------------------------------
ES2004c 8.8 7.2 1.3 0.4 4/4 72.4
ES2004b 8.9 7.3 1.0 0.7 4/4 75.6
TS3003c 13.3 11.1 0.9 1.3 4/4 72.2
IS1009d 13.8 7.9 2.1 3.8 4/4 74.8
TS3003b 16.2 5.9 1.6 8.8 4/4 73.0
TS3003a 19.0 16.6 0.8 1.6 4/4 77.0
EN2002b 20.4 16.0 1.7 2.8 4/4 75.9
TS3003d 20.5 14.7 1.9 3.9 4/4 72.1
IS1009c 22.1 6.8 2.0 13.4 4/4 74.4
EN2002c 23.2 16.6 1.9 4.7 4/3 69.9
IS1009a 23.3 7.9 2.7 12.7 4/4 81.8
EN2002a 24.4 19.6 1.1 3.7 4/4 73.2
IS1009b 25.7 4.8 1.7 19.2 4/4 74.9
ES2004d 27.7 15.1 1.5 11.2 4/4 73.1
EN2002d 27.9 21.9 2.2 3.7 4/4 73.2
ES2004a 35.6 13.4 19.1 3.0 4/4 78.1
----------------------------------------------------------------------
AVERAGE 20.7 12.1 2.7 5.9 - 74.5
======================================================================
CharsiuG2P ByT5 encoder-decoder model converted to CoreML for multilingual grapheme-to-phoneme conversion. Used by Kokoro TTS for non-English phonemization.
Model: FluidInference/charsiu-g2p-byt5-coreml
Hardware: Apple M2, 2022, macOS 26
swift run -c release fluidaudiocli g2p-benchmark --data-dir /path/to/CharsiuG2P/data/test| Language | PER | WER | ms/word |
|---|---|---|---|
| Spanish | 0.1% | 0.8% | 32.6 |
| French | 0.8% | 2.0% | 26.5 |
| Italian | 2.8% | 20.0% | 20.9 |
| Hindi | 4.5% | 21.4% | 45.4 |
| Japanese | 10.5% | 23.8% | 31.7 |
| Portuguese (BR) | 8.9% | 43.2% | 24.0 |
| British English | 13.6% | 29.4% | 34.0 |
| American English | 19.0% | 38.8% | 28.2 |
| Chinese | 86.2%* | 95.0%* | 53.9 |
| Average | 16.3% | 30.5% | 33.0 |
*Chinese PER is inflated due to tone notation mismatch between model output and reference data (tone contour marks vs model format), not a model accuracy issue.
- PER (Phoneme Error Rate): Character-level Levenshtein distance / reference length, stress marks stripped
- WER (Word Error Rate): Fraction of words with any phoneme error
Both the English BART G2P and multilingual ByT5 G2P models run fastest on CPU-only due to GPU/ANE dispatch overhead on small autoregressive decoder steps.
Multilingual G2P (ByT5)
| Compute Units | ms/word |
|---|---|
| cpuOnly | 38.7 |
| cpuAndGPU | 94.7 |
| all (ANE+GPU+CPU) | 95.2 |
English G2P (BART)
| Compute Units | ms/word |
|---|---|
| cpuOnly | 13.0 |
| all (ANE+GPU+CPU) | 17.3 |
| cpuAndGPU | 23.4 |
Parakeet CTC 0.6B zh-CN model converted to CoreML for on-device Mandarin Chinese transcription.
⚠️ Experimental Feature: This is an early preview of Mandarin Chinese ASR support. The API and performance characteristics may change in future releases.
Model: FluidInference/parakeet-ctc-0.6b-zh-cn-coreml
Hardware: Apple M2, 2022, macOS 26
Full benchmark on the complete THCHS-30 test set — 2,495 utterances (250 unique sentences × 10 speakers) from the THCHS-30 corpus.
Dataset: FluidInference/THCHS-30-tests
swift run -c release fluidaudiocli ctc-zh-cn-benchmark --auto-download| Metric | int8 encoder (0.55 GB) |
|---|---|
| Mean CER | 8.23% |
| Median CER | 6.45% |
| CER = 0% (perfect) | 435 (17.4%) |
| CER < 5% | 947 (38.0%) |
| CER < 10% | 1,674 (67.1%) |
| CER < 20% | 2,325 (93.2%) |
| Mean Latency | 614 ms |
| Mean RTFx | 14.83x |
Error analysis from the 100 highest-CER samples (out of the full 2,495) identified 862 substitution errors. The dominant patterns:
- Homophones / near-homophones: acoustically similar syllables (e.g. 呢/了, 了/的) account for the majority of substitutions — unavoidable without a language model
- Digit representation: the model may output Arabic digits (1, 5, 2011) when references use Chinese characters (一五, 二零一一); the benchmark normalizer converts digits before scoring to avoid penalizing this
- Sentence-final particles: 了/的/呢/吧 are frequently confused, contributing a disproportionate share of errors given their high occurrence
Beam search does not improve CER for this model without a language model. Greedy decoding (beam width 1) is recommended.
- Greedy decoding is sufficient for production use at this CER level.
- For applications requiring <8% CER, a character-level language model would be needed.
- Int8 encoder (0.55 GB) performs on par with FP32 (1.1 GB).
Parakeet TDT 0.6B Japanese model converted to CoreML for on-device Japanese transcription. Hybrid architecture using CTC preprocessor/encoder with TDT v2 decoder/joint (workaround for CoreML conversion bug).
Model: FluidInference/parakeet-ctc-0.6b-ja-coreml
Hardware: Apple M2, 2022, macOS 26
Full benchmark on the complete JSUT-basic5000 dataset — 5,000 utterances from a single Japanese speaker.
Dataset: JSUT-basic5000
swift run -c release fluidaudiocli ja-benchmark --decoder tdt --dataset jsut --samples 5000 --auto-download| Metric | TDT Decoder |
|---|---|
| Mean CER | 6.88% |
| Median CER | 4.08% |
| CER < 5% | 2,683 (53.7%) |
| CER < 10% | 3,549 (71.0%) |
| CER < 20% | 4,556 (91.1%) |
| Mean Latency | 208.8 ms |
| Mean RTFx | 28.9x |
Note: CER calculation includes number normalization (full-width digits → half-width, kanji numbers → Arabic) matching NVIDIA's evaluation methodology. NVIDIA reports 6.4% CER for the same model on JSUT-basic5000.