Update README.md

README.md

@@ -10,10 +10,11 @@ base_model:
 - facebook/wav2vec2-large-960h-lv60-self
 ---
 
+
 # 🔍 **AntiDeepfake**
 
-The **AntiDeepfake** project provides a series of powerful self-supervised learning (SSL) models crafted to **detect deepfake speech** with state-of-the-art accuracy.
 
+The **AntiDeepfake** project provides a series of powerful foundation models post-trained for **deepfake detection**. The AntiDeepfake model can be used for feature extraction for deepfake detection in a zero-shot manner, or it may be further fine-tuned and optimized for a specific database or deepfake-related task.
 
 # 🤖 Available Models
 
@@ -195,11 +196,24 @@ Results shown below can be reproduced using scripts provided in our [GitHub repo
 | FakeOrReal           | 1.000   | 0.992    | 0.994     | 0.989  | 0.991    | 0.005 | 0.011 | 0.63 @ 0.3727        |
 | FakeOrReal-norm      | 0.999   | 0.986    | 0.975     | 0.997  | 0.986    | 0.025 | 0.003 | 0.97 @ 0.7975        |
 | In-the-Wild          | 0.997   | 0.976    | 0.991     | 0.970  | 0.980    | 0.015 | 0.030 | 1.91 @ 0.3240        |
-| WildSVDD-TEST_A      | 0.739   | 0.663    | 0.616     | 0.895  | 0.729    | 0.577 | 0.105 | 34.85 @ 0.9942       |
-| WildSVDD-TEST_B      | 0.654   | 0.696    | 0.740     | 0.836  | 0.785    | 0.583 | 0.164 | 41.67 @ 0.9842       |
+| Deepfake-Eval-2024  | 0.741   | 0.714    | 0.720     | 0.920  | 0.808    | 0.675 | 0.081 | 33.38 @ 0.9995       |
+
+You can also fine-tune this model on a specific database, the corresponding code is provided in our [GitHub repository](https://github.com/nii-yamagishilab/AntiDeepfake). Fine-tuning will follow a similar process to training a new model, except that model weights will be initialized as AntiDeepfake checkpoints.
+
+Below are the evaluation results of this model fine-tuned on the Deepfake-Eval-2024 training set and tested on its corresponding test set (as shown in the previous table):
+
+| Test Input Length | ROC AUC | Accuracy | Precision | Recall | F1-score | FPR    | FNR    | EER (%) @ Threshold |
+|-------------------|---------|----------|-----------|--------|----------|--------|--------|----------------------|
+| 4s                | 0.8776  | 0.8147   | 0.8126    | 0.9315 | 0.8680   | 0.4060 | 0.0685 | 19.56 @ 0.9238       |
+| 10s               | 0.9362  | 0.8831   | 0.8914    | 0.9346 | 0.9125   | 0.2135 | 0.0654 | 12.10 @ 0.8521       |
+| 13s               | 0.9440  | 0.8977   | 0.9114    | 0.9337 | 0.9224   | 0.1697 | 0.0663 | 10.94 @ 0.8127       |
+| 30s               | 0.9495  | 0.8975   | 0.9310    | 0.9088 | 0.9198   | 0.1233 | 0.0912 | 10.52 @ 0.6061       |
+| 50s               | 0.9414  | 0.8792   | 0.9348    | 0.8707 | 0.9016   | 0.1060 | 0.1293 | 11.37 @ 0.4168       |
+
+
 
 # **Training Set**
-Below is a breakdown of the training set used for fine-tuning.
+Below is a breakdown of the training set used for post-training of speech SSL models.
 
 | 📚 Database             | 🌍 Language         | ✅ Genuine (hrs) | ❌ Fake (hrs) |
 |----------------------|------------------|----------------|-------------|
@@ -244,5 +258,6 @@ This project is based on results obtained from project JPNP22007, commissioned b
 
 It is also partially supported by the following grants from the Japan Science and Technology Agency (JST):
 - AIP Acceleration Research (Grant No. JPMJCR24U3)
-- CREST (Grant No. JPMJCR20D3)
-- PRESTO (Grant No. JPMJPR23P9)
+- PRESTO (Grant No. JPMJPR23P9)
+
+This study was carried out using the TSUBAME4.0 supercomputer at Institute of Science Tokyo.