From cad4b10bf6aeae01e0fb872e051749c878aedff8 Mon Sep 17 00:00:00 2001 From: Eason WaveKat Date: Fri, 27 Mar 2026 08:08:12 +1300 Subject: [PATCH] docs: fix incorrect benchmark claims in README Move "best F1" claim from FireRedVAD to Silero (which actually has the highest F1 at 0.938) and remove unsupported AUC-ROC claim from FireRedVAD. Co-Authored-By: Claude Opus 4.6 (1M context) --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index f08647d..592f9ae 100644 --- a/README.md +++ b/README.md @@ -75,7 +75,7 @@ let result = vad.process(&samples, 16000).unwrap(); // 0.0 or 1.0 ### Silero -Neural network (LSTM) via ONNX Runtime. Returns continuous probability, higher accuracy than WebRTC. Only supports 8kHz and 16kHz. +Neural network (LSTM) via ONNX Runtime. Returns continuous probability, best overall F1 across benchmarks. Only supports 8kHz and 16kHz. ```rust use wavekat_vad::VoiceActivityDetector; @@ -104,7 +104,7 @@ let probability = vad.process(&samples, 16000).unwrap(); // 0.0–1.0 ### FireRedVAD -Xiaohongshu's FireRedVAD using a DFSMN architecture with pure Rust FBank preprocessing. Returns continuous probability, 16kHz only. Best overall F1 and AUC-ROC across benchmarks. +Xiaohongshu's FireRedVAD using a DFSMN architecture with pure Rust FBank preprocessing. Returns continuous probability, 16kHz only. ```rust use wavekat_vad::VoiceActivityDetector;