image2speech: A Python repository from zguesmi

Description

image2speech is an Ethereum ready dapp that combines the functionalities of two other dapps: image2text and text2speech. It applies google's tesseract-OCR engine to extract text from images and uses mimic tts engine to convert text to speech and save them in wav format.

Usage

Bring your images together in a folder (exp: DATADIR) and add an input-config.yml file in the same folder. This file contains configuration for each image it should respect the following format:

<imagename 1>:
    language: <lang>
    voice: <voice name>
    latency: <value>

<imagename 2>:
    language: <lang>
    voice: <voice name>
    latency: <value>

You can keep the language's section empty but this may affect the performance of the extraction process. If voice name and latency are not defined the app will default to "ap" voice and 1.0 latency.

In the app/app-config.yml file, change the datadir parameter to the path of your folder (DATADIR for our example) and run the script.

$ git clone https://github.com/Zied-Guesmi/image2speech.git && cd image2speech/
$ sed -i "s/\/iexec/<path/to/datadir/folder>/" app/app-config.yml
$ python3 app/src/app.py

You shoud find your text/sound files in the DATADIR/out/ folder.

Supported languages

English (for now).

Supported image types

Tested extensions: jpeg, bmp, png
Those extensions are accepted but were not tested yet: pbm, pgm, ppm, tiff, rast, xbm

Voices

Male: ap, awb_time, awb, kal, kal16, rms.
Female: slt_hts, slt.
See voice samples here.

Dependencies

Docker deployment

After installing docker and preparing your DATADIR folder, just grab the docker image from dockerhub and run it:

$ docker run -v abs/path/to/datadir:/iexec/ ziedguesmi/image2speech

Or you can build your own image from dockerfile:

# clone the dapp
$ git clone https://github.com/Zied-Guesmi/image2speech.git && cd image2speech/ 

# build the docker image
$ docker build -t image2speech .

# run the container
$ docker run -v abs/path/to/datadir:/iexec/ image2speech

Installation

Install system dependencies:

# ubuntu as an example
$ apt-get update && apt-get install -y \
    automake \
    gcc \
    git \
    libasound2-dev \
    libsm6 \
    libtesseract-dev \
    libtool \
    make \
    pkg-config \
    python3 \
    python3-pip \
    tesseract-ocr \
    tesseract-ocr-ara \
    tesseract-ocr-eng \
    tesseract-ocr-fra \
    tesseract-ocr-spa \
    tesseract-ocr-deu \
    tesseract-ocr-chi-sim \
    tesseract-ocr-ita \
    tesseract-ocr-jpn \
    tesseract-ocr-por \
    tesseract-ocr-rus \
    tesseract-ocr-tur \
    tesseract-ocr-kor \
    unzip \
    wget

Install mimic tts engine

$ git clone https://github.com/MycroftAI/mimic.git && cd mimic
$ ./dependencies.sh --prefix="/usr/local"
$ ./autogen.sh
$ ./configure --prefix="/usr/local"
$ make

Clone the app:

$ git clone https://github.com/Zied-Guesmi/image2speech.git && cd image2speech/

Install python depedencies:

$ pip3 install -r app/requirements.txt

TODO

Support more languages.
Add more mimic voices.
Reduce mimic docker image size.

Authors

Zied Guesmi

License

This project is licensed under the MIT License - see the LICENSE file for details

zguesmi/image2speech