generate_audio.py

#!/usr/bin/env python3

import json
from pathlib import Path
from gtts import gTTS
import pandas as pd


def create_audio_directory():
    audio_dir = Path("audio")
    if not audio_dir.exists():
        audio_dir.mkdir(parents=True)
    return audio_dir


def generate_mp3_from_text(text, output_path, lang="vi"):
    try:
        tts = gTTS(text=text, lang=lang, slow=False)
        tts.save(output_path)
        return True
    except Exception as e:
        print(f"Error generating audio for '{text}': {e}")
        return False


def process_dataset(jsonl_file="vietnamese_ipa_dataset.jsonl"):
    create_audio_directory()

    # Read JSONL file
    records = []
    with open(jsonl_file, "r", encoding="utf-8") as f:
        for line in f:
            records.append(json.loads(line))

    df = pd.DataFrame(records)

    total = len(df)
    success = 0
    failed = []

    print(f"Processing {total} words from {jsonl_file}...")
    print("-" * 50)

    for idx, row in df.iterrows():
        word = row["word"]
        ipa = row["ipa"]
        audio_path = row["audio"]["path"] if isinstance(row["audio"], dict) else row["audio"]

        full_path = Path(audio_path)

        print(f"[{idx + 1}/{total}] Generating: {word} [{ipa}]")

        if generate_mp3_from_text(word, str(full_path)):
            success += 1
            print(f"  ✓ Saved to: {full_path}")
        else:
            failed.append(word)
            print("  ✗ Failed to generate audio")

    print("-" * 50)
    print("\nResults:")
    print(f"  Successfully generated: {success}/{total}")
    if failed:
        print(f"  Failed words: {', '.join(failed)}")

    return success, failed


def add_new_word(word, ipa, jsonl_file="vietnamese_ipa_dataset.jsonl"):
    audio_filename = f"audio/{word.replace(' ', '_')}.mp3"

    # Create new record
    record = {
        "word": word,
        "ipa": ipa,
        "audio": {
            "path": audio_filename,
            "sampling_rate": 22050
        }
    }

    # Append to JSONL file
    with open(jsonl_file, "a", encoding="utf-8") as f:
        f.write(json.dumps(record, ensure_ascii=False) + "\n")

    create_audio_directory()
    if generate_mp3_from_text(word, audio_filename):
        print(f"Added '{word}' with IPA [{ipa}] and generated audio")
        return True
    else:
        print(f"Added '{word}' to CSV but failed to generate audio")
        return False


def display_dataset(jsonl_file="vietnamese_ipa_dataset.jsonl"):
    # Read JSONL file
    records = []
    with open(jsonl_file, "r", encoding="utf-8") as f:
        for line in f:
            records.append(json.loads(line))

    df = pd.DataFrame(records)
    print("\nVietnamese IPA Dataset:")
    print("-" * 70)
    print(f"{'Word':<20} {'IPA':<30} {'Audio File':<20}")
    print("-" * 70)

    for _, row in df.iterrows():
        audio_path = row["audio"]["path"] if isinstance(row["audio"], dict) else row["audio"]
        print(f"{row['word']:<20} {row['ipa']:<30} {audio_path:<20}")

    print("-" * 70)
    print(f"Total entries: {len(df)}")


if __name__ == "__main__":
    import sys

    if len(sys.argv) > 1:
        if sys.argv[1] == "generate":
            process_dataset()
        elif sys.argv[1] == "display":
            display_dataset()
        elif sys.argv[1] == "add" and len(sys.argv) >= 4:
            word = sys.argv[2]
            ipa = sys.argv[3]
            add_new_word(word, ipa)
        else:
            print("Usage:")
            print("  python generate_audio.py generate    # Generate all MP3 files")
            print("  python generate_audio.py display     # Display dataset")
            print("  python generate_audio.py add <word> <ipa>  # Add new word")
    else:
        print("Generating audio files for Vietnamese IPA dataset...")
        process_dataset()
        print("\nDataset overview:")
        display_dataset()