prepare_model.py

import json
import os
import subprocess

# Load the data from the provided dictionary
data = {
    "VCTK_p283": {
        "age": 24,
        "gender": "F",
        "accents": "Irish",
        "region": "Cork"
    },
    "VCTK_p266": {
        "age": 22,
        "gender": "F",
        "accents": "Irish",
        "region": "Athlone"
    },
    "VCTK_p288": {
        "age": 22,
        "gender": "F",
        "accents": "Irish",
        "region": "Dublin"
    },
    "VCTK_p295": {
        "age": 23,
        "gender": "F",
        "accents": "Irish",
        "region": "Dublin"
    },
    "VCTK_p293": {
        "age": 22,
        "gender": "F",
        "accents": "NorthernIrish",
        "region": "Belfast"
    },
    "VCTK_p238": {
        "age": 22,
        "gender": "F",
        "accents": "NorthernIrish",
        "region": "Belfast"
    },
    "VCTK_p261": {
        "age": 26,
        "gender": "F",
        "accents": "NorthernIrish",
        "region": "Belfast"
    },
    "VCTK_p351": {
        "age": 21,
        "gender": "F",
        "accents": "NorthernIrish",
        "region": "Derry"
    },
    "VCTK_p249": {
        "age": 22,
        "gender": "F",
        "accents": "Scottish",
        "region": "Aberdeen"
    },
    "VCTK_p234": {
        "age": 22,
        "gender": "F",
        "accents": "Scottish",
        "region": "West Dumfries"
    },
    "VCTK_p262": {
        "age": 23,
        "gender": "F",
        "accents": "Scottish",
        "region": "Edinburgh"
    },
    "VCTK_p264": {
        "age": 23,
        "gender": "F",
        "accents": "Scottish",
        "region": "West Lothian"
    },
    "VCTK_p265": {
        "age": 23,
        "gender": "F",
        "accents": "Scottish",
        "region": "Ross"
    },
    "VCTK_p253": {
        "age": 22,
        "gender": "F",
        "accents": "Welsh",
        "region": "Cardiff"
    },
    "VCTK_p313": {
        "age": 24,
        "gender": "F",
        "accents": "Irish",
        "region": "County Down"
    },
    "VCTK_p340": {
        "age": 18,
        "gender": "F",
        "accents": "Irish",
        "region": "Dublin"
    },
    "VCTK_p335": {
        "age": 25,
        "gender": "F",
        "accents": "NewZealand",
        "region": "English"
    },
    "VCTK_p280": {
        "age": 25,
        "gender": "F",
        "accents": "France",
        "region": "France"
    }
}

# Convert the data to JSON format
json_data = json.dumps(data, indent=2)

# Save the JSON data to a file
with open('speakers-log.json', 'w') as file:
    file.write(json_data)

# Run the TTS command to get the speaker indices
command = "tts --model_path checkpoint_85000.pth --config_path config.json --list_speaker_idxs | grep -vE '^(\s*\||\s*>|\s*$)'"
output = subprocess.check_output(command, shell=True, text=True)

# Parse the JSON output into a Python dictionary
speaker_indices = eval(output)

# Load the speaker IDs from speakers.json
with open('speakers-log.json', 'r') as file:
    speaker_ids = json.load(file)

for speaker_idx in speaker_indices:
    # # Remove the 'VCTK_' prefix
    speaker_id = speaker_idx
    # speaker_id = speaker_idx.replace('VCTK_', '')

    # Lookup the speaker ID in the loaded speaker IDs
    if speaker_id in speaker_ids:
        speaker_id_json = speaker_ids[speaker_id]
    else:
        continue

    # # Generate the TTS command to create the audio file
    text = f"Hello, I am from {speaker_id_json['region']}. I hope that you will select my voice for your project. Thank you."
    # # make samples directory if it doesn't exist
    if not os.path.exists("samples"):
        os.makedirs("samples")

    out_path = f"samples/{speaker_id}.wav"
    tts_command = f"tts --text \"{text}\" --model_path checkpoint_85000.pth --language_idx en --config_path config.json --speaker_idx \"{speaker_id}\" --out_path {out_path}"

    # Execute the TTS command
    os.system(tts_command)