Create an audio program from a text file containing English sentences
Project description
apg (audio_program_generator)
Generates an audio program from text, with option to mix in background sound
Possible use cases:
- make your own yoga or qi gong routine
- play a list of daily affirmations
- meditate to a mantra with drone music in the background
- create an audio book
- read a kid a bedtime story without actually having to do the reading
Prerequisites
- Python (3.7+) [note to mac users: your system may be using Python 2.7 by default. To find out, issue the command
python --version
. If your system shows anything less than 3.7, make sure you create a virtual environment before installing this package (see Installation section below)] - pip (option 1)
- git + poetry (option 2)
- Local installation of ffmpeg
Installation & Execution
With pip
:
- Create a virtual environment and activate it:
python -m venv venv
source ./venv/bin/activate
- Install the package:
pip install audio-program-generator
- Once this is done, you will have an "apg" executable available in your terminal. You can type
apg
for basic help, orapg --help
for full instructions. - Deactivate your virtual environment when finished:
deactivate
With poetry
:
- Clone the repo and cd into the directory:
git clone https://github.com/jeffwright13/audio_program_generator.git
cd audio_program_generator
- Install the dependencies using poetry, and activate the virtual environment:
poetry install --no-dev
poetry shell
- Once this is done, you will have an "apg" executable available in your terminal. You can type
apg
for basic help, orapg --help
for full instructions. - Exit the poetry virtual environment when finished:
exit
With docker
:
- Clone the repo and cd into the directory:
git clone https://github.com/jeffwright13/audio_program_generator.git
cd audio_program_generator
- Execute
make build
(builds Docker image) - Execute
make run
(runs container and enters inpoetry
shell, ready to runapg
executable - Results from
make run
will be available locally in the/apgfiles
folder, even after the container is stopped
With flask
:
- There is a sister project that wraps the apg module in a bare-bones Flask app. This can be hosted locally, or in a cloud provider such as Heroku, Digital Ocean, or AWS. This method is considered experimental at the moment, and is not officially supported.
Usage
Assumes you are using the provided apg
command line interface, installed with one of the methods above
- Populate a semicolon-separated text file with plain-text phrases, each followed by an inter-phrase duration (see example below). Each line of the file is comprised of:
- a phrase to be spoken (in English)
- a semicolon
- a silence duration (in seconds)
- Provide a sound file for background sound (optional)
- Execute the command in your terminal:
apg [options] <phrase_file> [sound_file]
The script will generate and save a single MP3 file. The base name of the MP3 file is the same as the specified input file. For example, if the script is given input file "phrases.txt", the output file will be "phrases.mp3". It will be saved to the same folder that the input text file was taken from.
The optional [sound_file]
parameter, when specified, is used to mix in background sounds/music. This parameter specifies the path/filename of the sound file to be mixed in with the speech generated from the phrase file. If the sound file is shorter in duration than the generated speech file, it will be looped. If it is longer, it will be truncated. The resulting background sound (looped or not) will be faded in and out to ensure a smooth transition (6 seconds at beginning and en). Currently, only .wav files are supported as inputs.
The --attenuation
option allows fine-tuning the background sound level so it doesn't drown out the generated speech.
The --slow
option generates each speech snippet is a slow-spoken style.
The --tld
option allows the user to select one of several regional 'accents' (English only). For accents, select one from the following list: ["com.au", "co.uk", "com", "ca", "co.in", "ie", "co.za"]
Specifying option --book-mode
creates a spoken-word program (with or without background soundfile). It does this by reading in a file that does not have inter-phrase durations inserted, as is normally the case. This feature is new and needs some tweaking. For now, just make sure your input file is pure text, and experiement with using a single line (with many sentences) as one paragraph vs. multiple lines, one per sentence. You wil notice a difference in how the 'speaker' pauses between phrases.
The CLI prints out a progress bar as the phrase file is converted into speech snippets. No progress bar is shown for the secondary mix step. There may be a significant delay in going from the end of the first stage (snippet generation) to the end of the second stage (mixing), primarily because of reading in the .wav file, which may be large. For this reason, you may want to select a sound file for mixing that is small (suggested <20MB). Otherwise, be prepared to wait. The progress bar may be disabled with the --no-progress-bar
option.
Example <phrase_file> format:
Phrase One;2
Phrase Two;5
Phrase Three;0
Example --book-mode
file format:
Here we have sentence number one. It's a lovely sentence, and deserves its own paragraph.
Here is a second paragraph, and this is sentence number one (again) in that paragraph. And this is sentence number two! Then shalt thou count to three - no more, no less. Three shall be the number thou shalt count, and the number of the counting shall be three. Four shalt thou not count, neither count thou two, excepting that thou then proceed to three. Five is right out. Once the number three, being the third number, be reached, then lobbest thou thy Holy Hand Grenade of Antioch towards thy foe, who, being naughty in my sight, shall snuff it.
Author:
Jeff Wright jeff.washcloth@gmail.com
Collaborators:
Bob Belderbos Erik OShaughnessy
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for audio_program_generator-1.7.0.0.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | 78d6d2ec66d679eac27d2f861915d8c526653950fd8a3f1a5d164c0977cec774 |
|
MD5 | 11f36292f04140b1da0d7c7bed8cbb1e |
|
BLAKE2b-256 | bed46c5f3c04b4b41e8e309bbe08b485edbe885a3b171ca900d19435a98c1c98 |
Hashes for audio_program_generator-1.7.0.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | afb7216ab30e5118aef16c287208046740d6e4d887546a0b0c8039ecf241783c |
|
MD5 | 7cd8ec72f03fbe0d2399d943c54864a9 |
|
BLAKE2b-256 | 0e83ec0759139d581c9d97e5abee69b7af4fabe9a2be6bd47e2bb070491a50d6 |