Add code for variable-length

.gitignore

@@ -154,4 +154,8 @@ configs/local/default.yaml
 .aim
 
 # Neptune logging
-.neptune
+.neptune
+
+# Notebooks
+notebooks/**/*.txt
+notebooks/**/*.csv

configs/experiment/aasistssl_multiview_conf-2.1.yaml

@@ -0,0 +1,47 @@
+# @package _global_
+
+# to execute this experiment run:
+# python train.py experiment=example
+
+defaults:
+  - override /data: asvspoof_multiview
+  - override /model: xlsr_aasist_multiview
+  - override /callbacks: default_loss
+  - override /trainer: default
+
+# all parameters below will be merged with parameters from default configurations set above
+# this allows you to overwrite only specified parameters
+
+tags: ["asvspoof_multiview", "xlsr_aasist_multiview"]
+
+seed: 1234
+
+trainer:
+  max_epochs: 100
+  gradient_clip_val: 0.0
+  accelerator: cuda
+
+model:
+  optimizer:
+    lr: 0.000001
+    weight_decay: 0.0001
+  net: null
+  scheduler: null
+  compile: true
+
+data:
+  batch_size: 14
+  num_workers: 8
+  pin_memory: true
+  args:
+    padding_type: repeat
+    random_start: False
+    algo: 5 # Optimised for 5th algorithm LA
+
+
+logger:
+  wandb:
+    tags: ${tags}
+    group: "asvspoof_multiview"
+  aim:
+    experiment: "asvspoof_multiview"

configs/model/xlsr_conformer_reproduce.yaml

@@ -0,0 +1,21 @@
+_target_: src.models.xlsr_conformer_reproduce_module.XLSRConformerTCMLitModule
+
+optimizer:
+  _target_: torch.optim.Adam
+  _partial_: true
+  lr: 0.000001
+  weight_decay: 0.0001
+
+scheduler: null
+
+args:
+  conformer:
+    emb_size: 144
+    heads: 4
+    kernel_size: 31
+    n_encoders: 4
+
+ssl_pretrained_path: ${oc.env:XLSR_PRETRAINED_MODEL_PATH}
+cross_entropy_weight: [0.1, 0.9] # weight for cross entropy loss 0.1 for spoof and 0.9 for bonafide
+# compile model for faster training with pytorch 2.0
+compile: false

notebooks/analyze.ipynb

@@ -0,0 +1,121 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Processing audio files:   0%|          | 0/31779 [00:00<?, ?it/s]/tmp/ipykernel_3393222/3850030267.py:28: FutureWarning: get_duration() keyword argument 'filename' has been renamed to 'path' in version 0.10.0.\n",
+      "\tThis alias will be removed in version 1.0.\n",
+      "  duration = librosa.get_duration(filename=audio_path)\n",
+      "Processing audio files: 100%|██████████| 31779/31779 [02:46<00:00, 190.63it/s]"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Results written to in_the_wild_durations.csv\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "import os\n",
+    "import librosa\n",
+    "import csv\n",
+    "from tqdm import tqdm\n",
+    "\n",
+    "def get_audio_durations(audio_folder, output_csv):\n",
+    "    \"\"\"\n",
+    "    Calculate duration of each audio file in the folder and write results to CSV\n",
+    "    \n",
+    "    Parameters:\n",
+    "        audio_folder (str): Path to folder containing audio files\n",
+    "        output_csv (str): Path to output CSV file\n",
+    "    \"\"\"\n",
+    "    \n",
+    "    # Get list of audio files\n",
+    "    audio_files = [f for f in os.listdir(audio_folder) if f.endswith(('.wav', '.mp3', '.flac'))]\n",
+    "    \n",
+    "    # Open CSV file to write results\n",
+    "    with open(output_csv, 'w', newline='') as csvfile:\n",
+    "        writer = csv.writer(csvfile)\n",
+    "        writer.writerow(['filename', 'duration'])  # Write header\n",
+    "        \n",
+    "        # Process each audio file with progress bar\n",
+    "        for audio_file in tqdm(audio_files, desc=\"Processing audio files\"):\n",
+    "            try:\n",
+    "                # Load audio file and get duration\n",
+    "                audio_path = os.path.join(audio_folder, audio_file)\n",
+    "                duration = librosa.get_duration(filename=audio_path)\n",
+    "                \n",
+    "                # Write result to CSV\n",
+    "                writer.writerow([audio_file, f\"{duration:.2f}\"])\n",
+    "                \n",
+    "            except Exception as e:\n",
+    "                print(f\"Error processing {audio_file}: {str(e)}\")\n",
+    "\n",
+    "if __name__ == \"__main__\":\n",
+    "    # Example usage\n",
+    "    audio_folder = \"/data/hungdx/Lightning-hydra/data/in_the_wild\"\n",
+    "    output_csv = \"in_the_wild_durations.csv\"\n",
+    "    \n",
+    "    get_audio_durations(audio_folder, output_csv)\n",
+    "    print(f\"Results written to {output_csv}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Average duration:  4.287989552849366\n"
+     ]
+    }
+   ],
+   "source": [
+    "import pandas as pd\n",
+    "\n",
+    "df = pd.read_csv('in_the_wild_durations.csv')\n",
+    "\n",
+    "print(\"Average duration: \", df['duration'].mean())\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "asvspoof5",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.19"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

notebooks/eval.ipynb

@@ -0,0 +1,79 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "eer: 9.326796296235376\tthreshold: -2.986328125\n",
+      "\n",
+      "0.09326796296235375\n"
+     ]
+    }
+   ],
+   "source": [
+    "import sys\n",
+    "import os.path\n",
+    "import numpy as np\n",
+    "import pandas\n",
+    "import eval_metrics_DF as em\n",
+    "\n",
+    "\n",
+    "def eval_to_score_file(score_file, cm_key_file):\n",
+    "    # CM key file is the metadata file that contains the ground truth labels for the eval set\n",
+    "    # score file is the output of the system that contains the scores for the eval set\n",
+    "    # phase is the phase of the eval set (dev or eval)\n",
+    "\n",
+    "    cm_data = pandas.read_csv(cm_key_file, sep=' ', header=None)\n",
+    "    submission_scores = pandas.read_csv(\n",
+    "        score_file, sep=' ', header=None, skipinitialspace=True)\n",
+    "    # check here for progress vs eval set\n",
+    "    cm_scores = submission_scores.merge(cm_data, left_on=0, right_on=0, how='inner')\n",
+    "    # cm_scores.head()\n",
+    "    #  0       1_x   1_y      2      3\n",
+    "    #  a.wav  1.234   eval   Music   spoof\n",
+    "    bona_cm = cm_scores[cm_scores['1_y'] == 'bonafide']['1_x'].values\n",
+    "    spoof_cm = cm_scores[cm_scores['1_y'] == 'spoof']['1_x'].values\n",
+    "\n",
+    "    eer_cm, th = em.compute_eer(bona_cm, spoof_cm)\n",
+    "    out_data = \"eer: {}\\tthreshold: {}\\n\".format(100*eer_cm, th)\n",
+    "    print(out_data)\n",
+    "    return eer_cm\n",
+    "\n",
+    "print(eval_to_score_file(\"/data/hungdx/Lightning-hydra/logs/eval/itw_xlsr_aasist_multiview_conf-2_epoch15_3s.txt\", \"/dataa/Datasets/in_the_wild.txt\"))"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "base",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}