scripts/01_mfcc_svm.py

from pathlib import Path
import numpy as np
import pandas as pd
from tqdm import tqdm
import librosa
import warnings
warnings.filterwarnings("ignore")

DATA_DIR = Path("output/linguawave")
SAMPLE_RATE = 16_000
DURATION = 10
CLASSES = ["id", "ms", "vi", "th", "en", "zh", "ar", "fr"]


def extract_features(fpath, sr=SAMPLE_RATE, n_mfcc=20):
    """Extract 40-dim MFCC feature vector (mean + std over time)."""
    y, _ = librosa.load(str(fpath), sr=sr, duration=DURATION)
    # Pad or trim to exact length
    target_len = sr * DURATION
    if len(y) < target_len:
        y = np.pad(y, (0, target_len - len(y)))
    else:
        y = y[:target_len]

    mfcc = librosa.feature.mfcc(y=y, sr=sr, n_mfcc=n_mfcc)  # (20, T)
    feats = np.concatenate([mfcc.mean(axis=1), mfcc.std(axis=1)])  # (40,)
    return feats


train_df = pd.read_csv(DATA_DIR / "train.csv")
test_df  = pd.read_csv(DATA_DIR / "test.csv")

from joblib import Parallel, delayed
Path("cache").mkdir(exist_ok=True)

def _ex(row_id):
    return extract_features(DATA_DIR / row_id)

def cached_extract(df, cache_name):
    cp = Path("cache") / cache_name
    if cp.exists():
        print(f"[cache] Loading {cache_name}"); return np.load(cp)
    feats = np.array(Parallel(n_jobs=-1, prefer="threads")(
        delayed(_ex)(r["id"]) for _, r in tqdm(df.iterrows(), total=len(df))))
    np.save(cp, feats); print(f"[cache] Saved {cache_name}"); return feats

print("Extracting train features ...")
X_train = cached_extract(train_df, "lw_01_train.npy")
y_train = train_df["label"].to_numpy()
print("X_train shape:", X_train.shape)

print("Extracting test features ...")
X_test = cached_extract(test_df, "lw_01_test.npy")
print("X_test shape:", X_test.shape)


from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder, StandardScaler

le = LabelEncoder()
le.fit(CLASSES)

y_enc = le.transform(y_train)

X_tr, X_val, y_tr, y_val = train_test_split(
    X_train, y_enc, test_size=0.15, random_state=42, stratify=y_enc
)
print(f"Train: {X_tr.shape[0]}  Val: {X_val.shape[0]}")


scaler = StandardScaler()
X_tr_s  = scaler.fit_transform(X_tr)
X_val_s = scaler.transform(X_val)
X_test_s = scaler.transform(X_test)


from sklearn.svm import SVC
from sklearn.metrics import f1_score, classification_report

clf = SVC(kernel="rbf", C=10, probability=True, random_state=42)
clf.fit(X_tr_s, y_tr)
print("SVM training complete.")


val_preds = clf.predict(X_val_s)
macro_f1 = f1_score(y_val, val_preds, average="macro")
print(f"Validation Macro F1: {macro_f1:.4f}")

print()
print(classification_report(
    y_val, val_preds,
    target_names=le.classes_,
))


test_preds_enc = clf.predict(X_test_s)
test_preds = le.inverse_transform(test_preds_enc)

sub = pd.DataFrame({"id": test_df["id"], "label": test_preds})
Path("submissions").mkdir(exist_ok=True)
sub.to_csv("submissions/sub_approach1_mfcc_svm.csv", index=False)
print("Saved submissions/sub_approach1_mfcc_svm.csv")
sub.head()