https://github.com/charlesaverill/ttsdg

A package to automate TTS data generation
https://github.com/charlesaverill/ttsdg

Last synced: about 6 hours ago
JSON representation

A package to automate TTS data generation

• Host: GitHub
• URL: https://github.com/charlesaverill/ttsdg
• Owner: CharlesAverill
• License: gpl-3.0
• Created: 2020-04-17T22:41:30.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-10-13T02:50:40.000Z (about 4 years ago)
• Last Synced: 2024-09-23T05:50:03.401Z (about 2 months ago)
• Language: Python
• Size: 488 KB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# TTSDG

TTSDG, or Text-To-Speech Data Generator, automates the simple-but-frustrating task of generating large amounts of TTS

data for tasks like machine learning. TTSDG contains an easy-to-use class that can generate text offline, in large 

batches, and with control over the system voices that you have installed. TTSDG randomizes volume, speed, and voice of 

each sample, and prevents duplicates from happening.

TTSDG utilizes [pyttsx3](https://pypi.org/project/pyttsx3/) and [pydub](https://pypi.org/project/pydub/) to generate the

audio and convert it into multiple formats. All pydub-supported formats are supported in TTSDG, like WAV, MP3, and AIFF.

# Installation

TTSDG is available through pip:

`python3 -m pip install ttsdg`

# Usage

```python

from ttsdg import TTSDG

for word in ["Apple", "Orange", "Banana"]:

    print(word)

    gen = TTSDG(verbose=True)

    gen.volume_range = [.5, 1.0]

    gen.wpm_range = [200, 300]

    

    gen.generate(word, 100, out_format="wav")

    

    # A bug in pyttsx3 will cause hangs on generation

    # in loops sometimes. del the generator at the end 

    # of the loop to solve this

    del gen

```

# Methods

- `set_volume_range(low, high, one)` - Set low or high volume bounds, or set `one` for a specific value.

- `set_wpm_range(low, high, one)` - Set low or high speed bounds (in words per minute), or set `one` for a specific 

value.

- `set_voices(voices)` - Only mess around with this if you know what you're doing. Sets the list of voices to choose

from, relies on what system voices you have installed. 

- `set_engine(engine)` - Only mess around with this if you know what you're doing. Sets the pyttsx3 Engine object,

useful for importing settings if your project already has an engine.

- `get_engine()` - Gets the pyttsx3 engine object.

- `generate(text, n, out_format)` - Generates `n` samples of the input `text`, saves them to 

`./(text)/(text)_index.(out_format)`