AirCaps A5S Async v1 Benchmark - ICSI SDM - cpWER: leaderboard

AirCaps A5S Async v1 Benchmark - ICSI SDM - cpWER — AirCaps A5S Async v1 author comparison, November2026. Same 16kHz mono16bit WAV full recordings and native API defaults, with only exact model, language and diarization set. English language is en; Mandarin datasets use their native language. Meta switches from file to realtime above10min with same model/mode. Other documented request limits use fewest equal chunks, two-minute overlap and reference-blind stitching. Failed recordings are empty transcripts/all deletions. References and hypotheses receive OpenASR normalization plus ellipsis removal and split-numeral joining, per speaker runs, then MeetEval minimum-permutation error. Nvidia is BF16 self-hosted1xH100, temperature0,top_p1,seed0,reasoningoff and its published benchmark prompt; NVIDIA generation-loop guardrail retained. Model runs and native request settings are attached. AirCaps accuracy runNovember6; peersNovember1–2. This campaign is distinct from VoiceArena or another evaluator using the same corpus. NVIDIA lists this corpus as training data; its score is retained as reported with this overlap explicitly recorded. Metric: cpWER (%). Source: https://github.com/aircaps/a5s-async-v1-benchmark

Metric: cpWER (%). Source: github.com. Saturation forecast: Rough model projection: around 2026. 9 models tracked. Updated 16 Nov 2026.

Top models

#ModelScoreOverall rank
1Nemotron 3.5 Transcribe10.04
2AirCaps A5S Async v114.63
3AssemblyAI Universal-3.5 Pro24.06
4Muse Voice Transcribe 1.028.48
5Scribe v236.91
6MAI-Transcribe-237.6
7Deepgram Nova 341.67
8Grok Voice Transcribe 2.042.2
9Gemini 3.5 Transcribe57.9

Interactive version: theaggregate.ai/benchmark?slug=aircaps-a5s-async-v1-benchmark-icsi-sdm-cpwer · How It Works · Data refreshed daily, snapshot 2026-11-16.