Add CER benchmark (plot + table); mark small fully trained

README.md

@@ -31,6 +31,25 @@ input (8-16 kHz telephony) --16k--> [FE: 24-layer w2v-BERT 2.0 + LoRA] --feature
 The **FE LoRA adapter is merged into the base weights at load time**, so inference needs **no `peft`** —
 just `transformers` + `descript-audio-codec`.
 
+## Benchmark — CER
+
+Character Error Rate (Whisper large-v3, **lower is better**) on a held-out **private call-centre
+test set**. CallEnhancer restores telephony-degraded speech so an ASR reads it far more accurately
+than the raw input or general-purpose speech enhancers.
+
+![CER on a private call-centre test set](assets/callcentre_cer.png)
+
+| model | CER % ↓ |
+|---|---:|
+| degraded (original) | 71.48 |
+| voicefixer | 58.27 |
+| resemble-enhance | 54.96 |
+| sidon-v0.1 | 41.06 |
+| **CallEnhancer-small** (this repo, open source) | **38.91** |
+| **CallEnhancer-base** | **19.01** |
+
+This repository hosts **CallEnhancer-small** (open source); **CallEnhancer-base** is the larger model.
+
 ## Quick start — infer from the HF checkpoint
 
 ```bash
@@ -52,8 +71,7 @@ cd CallEnhancer && python infer_callcentre.py \
 
 Prefer Python (load weights from the Hub with `hf_hub_download`)? See **[Python](#python-pull-weights-from-the-hub)** below.
 
-> **Status:** the decoder is still training and these checkpoints are refreshed periodically —
-> quality keeps improving. It already restores real 8 kHz call-centre audio well.
+> **Status:** CallEnhancer-small is fully trained and restores real 8 kHz call-centre audio well.
 
 ## Files