prepare_model.py

import json
import os
import subprocess

# Load the data from the provided dictionary
data = {
  "VCTK_p294": {"age": 33, "gender": "F", "accents": "American", "region": "San Francisco", "comments": ""},
  "VCTK_p297": {"age": 20, "gender": "F", "accents": "American", "region": "New York", "comments": ""},
  "VCTK_p299": {"age": 25, "gender": "F", "accents": "American", "region": "California", "comments": ""},
  "VCTK_p300": {"age": 23, "gender": "F", "accents": "American", "region": "California", "comments": ""},
  "VCTK_p301": {"age": 23, "gender": "F", "accents": "American", "region": "North Carolina", "comments": ""},
  "VCTK_p305": {"age": 19, "gender": "F", "accents": "American", "region": "Philadelphia", "comments": ""},
  "VCTK_p306": {"age": 21, "gender": "F", "accents": "American", "region": "New York", "comments": ""},
  "VCTK_p308": {"age": 18, "gender": "F", "accents": "American", "region": "Alabama", "comments": ""},
  "VCTK_p310": {"age": 21, "gender": "F", "accents": "American", "region": "Tennessee", "comments": ""},
  "VCTK_p318": {"age": 32, "gender": "F", "accents": "American", "region": "Napa", "comments": ""},
  "VCTK_p329": {"age": 23, "gender": "F", "accents": "American", "region": "America", "comments": ""},
  "VCTK_p330": {"age": 26, "gender": "F", "accents": "American", "region": "America", "comments": ""},
  "VCTK_p333": {"age": 19, "gender": "F", "accents": "American", "region": "Indiana", "comments": ""},
  "VCTK_p361": {"age": 19, "gender": "F", "accents": "American", "region": "New Jersey", "comments": ""},
  "VCTK_p362": {"age": 29, "gender": "F", "accents": "American", "region": "America", "comments": ""},
  "VCTK_p339": {"age": 21, "gender": "F", "accents": "American", "region": "Pennsylvania", "comments": ""},
  "VCTK_p341": {"age": 26, "gender": "F", "accents": "American", "region": "Ohio", "comments": ""}
}


# 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)