bangla draft

data/banglaser/README.md

@@ -0,0 +1,14 @@
+# Nkululeko pre-processing for Bangla SER dataset
+
+Download link: https://prod-dcd-datasets-cache-zipfiles.s3.eu-west-1.amazonaws.com/t9h6p943xy-5.zip
+Filename convention:  
+
+AA-BB-CC-DD-EE-FF-GG.wav
+
+```bash
+wget https://prod-dcd-datasets-cache-zipfiles.s3.eu-west-1.amazonaws.com/t9h6p943xy-5.zip
+
+```  
+
+Reference:  
+[1] 

data/banglaser/process_database.py

@@ -0,0 +1,90 @@
+# process_database.py for BanglaSER dataset
+# Path: data/banglaser/process_database.py
+
+"""
+Filename convention:
+AA-BB-CC-DD-EE-FF-GG.wav
+AA = mode, 03: audio only
+BB = script, 01: scripted
+CC = emotion, 01 = Happy, 02 = Sad, 03 = Angry, 04 = Surprise, 05 = Neutral
+DD = intensity, 01 = Normal, 02 = Strong
+EE = text
+01: It's twelve o'clock
+02: I knew something like this would happen.
+03: What kind of gift is this?
+FF = Repetition, 01, 02, 03
+GG = Actor ID, 01-34 (odd: male, even: female)
+"""
+
+import os
+
+import pandas as pd
+
+from nkululeko.utils.files import find_files
+
+
+def process_database(data_dir, output_dir):
+    # check if data_dir exists
+    if not os.path.isdir(data_dir):
+        print(f"ERROR: no such directory {data_dir}")
+        return
+    # create output dir if not exist
+    if not os.path.isdir(output_dir):
+        os.mkdir(output_dir)
+
+    # emotion mapping
+    emotion_map = {
+        "01": "happy",
+        "02": "sad",
+        "03": "angry",
+        "04": "surprise",
+        "05": "neutral",
+    }
+
+    # intensity mapping
+    intensity_map = {
+        "01": "normal",
+        "02": "strong",
+    }
+
+    # gender mapping, odd: male, even: female
+    gender_map = {
+        str(i): 'male' if i %
+        2 != 0 else 'female' for i in range(
+            1, 36)}
+
+    # load the data
+    wavs = find_files(data_dir, ext=["wav"], relative=True)
+
+    data = []
+
+    # loop over wavs
+    for wav in wavs:
+        # get emotion label from emotion_map
+        emotion = emotion_map[wav.split("-")[2]]
+        # get intensity label from intensity_map
+        intensity = intensity_map[wav.split("-")[3]]
+        # get gender label from gender_map
+        gender = gender_map[wav.split("-")[6]]
+        # get speaker ID
+        speaker = wav.split("-")[6]
+        # get script ID
+        text_id = wav.split("-")[4]
+        # get language
+        language = "bangla"
+
+        # add to data
+        data.append({
+            "wav": wav,
+            "emotion": emotion,
+            "intensity": intensity,
+            "gender": gender,
+            "speaker": speaker,
+            "text_id": text_id,
+            "language": language,
+        })
+
+    # convert to dataframe
+    df = pd.DataFrame(data)
+
+    # split to train,dev,test speaker independently, 8010:10

data/banglaser/t9h6p943xy-5

@@ -0,0 +1 @@
+/home/bagus/data/t9h6p943xy-5/