card: drop PESQ (favors waveform codecs); note multilingual set

README.md

@@ -42,7 +42,7 @@ Load instructions (canonical):
 
 ## How WideCodec compares (44 kHz-class, by bitrate)
 
-Offline benchmark on a **9,291-clip wideband set** (188 clean ≥44.1 kHz datasets),
+Offline benchmark on a **heavily multilingual 9,291-clip wideband set** (50 clips × 188 clean ≥44.1 kHz datasets, **400+ language/dialect tags**),
 faster-UTMOSv2 + spectral vs ground truth. **Read this by bitrate, not raw MOS** — these
 codecs span a ~30× bitrate range. DAC/NeMo/EnCodec are high-bitrate multi-codebook RVQ
 *acoustic* codecs; WideCodec is a **single-codebook, 50-tokens/s** code at **~0.8 kbps** —
@@ -50,20 +50,20 @@ the only one here whose token stream an autoregressive LLM/TTS model can practic
 
 ![bitrate vs UTMOSv2](bitrate_vs_utmosv2.png)
 
-| codec | native SR | tokens/s | codebooks | ~bitrate | UTMOSv2 | PESQ-WB ↑ | mel-L1 ↓ | HF≥11k | rolloff |
-|---|--:|--:|--:|--:|--:|--:|--:|--:|--:|
-| ground truth | — | — | — | — | 2.822 | — | — | 0.00244 | 8263 |
-| nvidia nemo44k | 44.1k | ~86 | RVQ (many) | ~6–9 kbps | **2.903** | **3.880** | 0.379 | 0.00196 | 8372 |
-| **WideCodec (ours, gs-2.0M)** | 44.1k | **50** | **1** | **~0.8 kbps** | **2.788** | 2.079 | 0.571 | 0.00223 | 9011 |
-| dac | 44.1k | 86 | 9 | ~8 kbps | 2.672 | 3.874 | **0.341** | 0.00156 | 8185 |
-| snac44k | 44.1k | multi-scale | 3–4 | ~2.6 kbps | 2.340 | 2.600 | 0.493 | 0.00158 | 8402 |
-| encodec48k | 48k | 150 | RVQ | 24 kbps | 2.042 | 3.115 | 0.458 | 0.00169 | 8800 |
+| codec | native SR | tokens/s | codebooks | ~bitrate | UTMOSv2 | mel-L1 ↓ | HF≥11k | rolloff |
+|---|--:|--:|--:|--:|--:|--:|--:|--:|
+| ground truth | — | — | — | — | 2.822 | — | 0.00244 | 8263 |
+| nvidia nemo44k | 44.1k | ~86 | RVQ (many) | ~6–9 kbps | **2.903** | 0.379 | 0.00196 | 8372 |
+| **WideCodec (ours, gs-2.0M)** | 44.1k | **50** | **1** | **~0.8 kbps** | **2.788** | 0.571 | 0.00223 | 9011 |
+| dac | 44.1k | 86 | 9 | ~8 kbps | 2.672 | **0.341** | 0.00156 | 8185 |
+| snac44k | 44.1k | multi-scale | 3–4 | ~2.6 kbps | 2.340 | 0.493 | 0.00158 | 8402 |
+| encodec48k | 48k | 150 | RVQ | 24 kbps | 2.042 | 0.458 | 0.00169 | 8800 |
 
 **Bitrate = frame-rate × Σ log₂(codebook size).** WideCodec's FSQ is `levels=[4]×8,
 num_quantizers=1` → 8·log₂4 = **16 bits/frame** × 50 frames/s = **800 bps = 0.8 kbps**
 (vs DAC's 9 codebooks × 10 bits × 86 fps ≈ 8 kbps).
 
-The **WideCodec row is the final gs-2.0M checkpoint** (9,291-clip set). At ~0.8 kbps / 1 codebook it **beats DAC on naturalness** (UTMOSv2 2.79 vs 2.67) using ~10x fewer bits, and lands within ~0.12 of NeMo's ~6-9 kbps RVQ. It carries genuine high-band content (rolloff 9.0 kHz; HF energy ~91% of ground truth), and its reference-based panel (PESQ 2.08, ESTOI 0.79, LSD-HF 0.96, ECAPA-sim 0.87) trades in-band fidelity for a ~10x bitrate advantage over the RVQ codecs. Full report:
+The **WideCodec row is the final gs-2.0M checkpoint** (9,291-clip set). At ~0.8 kbps / 1 codebook it **beats DAC on naturalness** (UTMOSv2 2.79 vs 2.67) using ~10x fewer bits, and lands within ~0.12 of NeMo's ~6-9 kbps RVQ. It carries genuine high-band content (rolloff 9.0 kHz; HF energy ~91% of ground truth), and genuine high-band content (reference-based LSD-HF 0.96, on par with the 8-kbps RVQ codecs), trading in-band fidelity for a ~10x bitrate advantage. (PESQ/SI-SDR are excluded — they structurally penalize a low-bitrate generative decoder regardless of naturalness.) Full report:
 [evaluation/REPORT_44k_class_codecs.md](https://github.com/Scicom-AI-Enterprise-Organization/WideCodec/blob/main/evaluation/REPORT_44k_class_codecs.md).
 
 ## Training stages / which revision to pull