How To Create Free AI Audio From Texts (Transcripts) for your Youtube Video using Python

What is Text-to-Speech (TTS) and Why Use It?
Text-to-Speech (TTS) converts written text into spoken words, ideal for video screencasts. It automates voiceovers, offering multilingual options, consistent narration, and varied voices (e.g., male, female, accents).

What is Espeak-NG?
Espeak-NG is an open-source speech synthesizer. It’s lightweight, fast, and works offline. While its voices are less natural than AI tools, it’s perfect for efficient, customizable TTS tasks.

Installing Espeak-NG
For Windows:

1. Download Espeak-NG from the official page.
2. Install it and add it to PATH.
3. Verify:

   espeak-ng --version
   

   .

For macOS:

1. Install Homebrew: brew install espeak.
2. Verify:

   espeak-ng --version
   

   .

For Linux (Ubuntu/Debian):

1. Run: sudo apt update && sudo apt install espeak-ng.
2. Verify:

   espeak-ng --version
   

   .

Testing Espeak-NG
Run:

espeak-ng "Welcome to vikaskbh.com"

Listen to the output to ensure proper installation.

Python in TTS Workflows
Python automates the TTS process:

1. Reads text from files.
2. Sends it to Espeak-NG.
3. Saves the audio as WAV or MP3.

Setting Up Python Libraries
Install Python 3 and pydub for audio conversion:

pip install pydub

Ensure FFmpeg is installed for pydub:

• Windows: Add FFmpeg to PATH (FFmpeg Download).
• macOS/Linux:

  brew install ffmpeg
  

  .

Python Script for TTS

import os
from pydub import AudioSegment

def text_to_speech(input_file, output_file):
    with open(input_file, "r") as file:
        text = file.read()

    temp_wav = "temp.wav"
    os.system(f'espeak-ng "{text}" -w {temp_wav}')

    audio = AudioSegment.from_wav(temp_wav)
    audio.export(output_file, format="mp3")
    os.remove(temp_wav)
    print(f"Audio saved as {output_file}")

text_to_speech("script.txt", "output.mp3")

Save your text in script.txt and run the script to generate audio.

Using Generated Audio in Video Screencasts

1. Open your video editor (e.g., Premiere Pro, Camtasia).
2. Import the MP3 file.
3. Sync the audio with video content on the timeline.

Customizing Voice with Espeak-NG
Adjust pitch, speed, and volume using these flags:

• -p [pitch] (0-99, default 50).
• -s [speed] (default 175 WPM).
• -a [amplitude] (0-200, default 100).

Example:

speak-ng -p 30 -s 150 -a 200 "Deeper and slower voice" -w output.wav

Modify these options in Python scripts as needed.

Preparing Reference Audio for Voice Cloning
Use voice cloning models in Coqui TTS. Prepare speaker_wav files:

1. Record a 5-10 second sample.
2. Convert to the required format with FFmpeg:

   ffmpeg -i speaker1.wav -ar 16000 -ac 1 speaker1_converted.wav
   

3. Use the file in TTS: from TTS.api import TTS

from TTS.api import TTS
# Load the TTS model
current_model = "tts_models/multilingual/multi-dataset/xtts_v2"
tts = TTS(model_name=current_model, gpu=False)
# Read text from file
with open("./transcript.txt", "r") as file:
    text = file.read()
tts.tts_to_file(text=text, file_path="output.wav", language = 'en',  speaker_wav="cloned_voice.wav")

Troubleshooting Common Errors

1. Espeak-NG not found:
   • Add Espeak-NG to PATH.
2. FFmpeg not found:
   • Install FFmpeg and ensure it’s in PATH.
3. Audio quality issues:
   • Experiment with pitch, speed, and amplitude settings.

So, This simplifies TTS workflows for creating professional voiceovers in your video screencasts. Tackle common issues, customize voice outputs, and explore advanced cloning for multilingual and natural-sounding audio.