transcribe

Requirements

• Python3
• Tensorflow version 1.1
• Seq2seq: https://google.github.io/seq2seq/getting_started/
• Pandas

If you use Anaconda, all dependencies can be installed into separate Anaconda environment:

conda create -n python36_tf11 python=3.6
source activate python36_tf11
export Q=https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-1.1.0-py3-none-any.whl
pip install --ignore-installed --upgrade $Q
conda install pandas

git clone https://github.com/google/seq2seq.git
cd seq2seq
pip install -e .

Usage

To transcribe phrases in test_in1.txt using default model run

python3 transcribe_seq2seq.py

Transcriptions will be written to test_out1.txt.

Input, output files and model can be specified explicitly:

python3 transcribe_seq2seq.py -fin test_in1.txt -fout test_out1.txt -model word3stress_bahdanau/model.ckpt-14850