Skip to content

Latest commit

 

History

History
15 lines (9 loc) · 729 Bytes

README.md

File metadata and controls

15 lines (9 loc) · 729 Bytes

CNN for discourse_prediction in a storytelling style speech

Python code for the paper "Sentence Based Discourse Classification for Hindi Story Text-to-Speech (TTS) System"

Runs the model on Hindi Story discourse dataset (Desceptive, Narrative and Dialogue). Please cite the original paper when using the data.

Steps for running the model

1- python process_data.py

2- # Any one of them

THEANO_FLAGS=mode=FAST_RUN,device=cpu,floatX=float32 python convolution_net_sentence.py -nonstatic -rand THEANO_FLAGS=mode=FAST_RUN,device=cpu,floatX=float32 python convolution_net_sentence.py -static -word2vec THEANO_FLAGS=mode=FAST_RUN,device=cpu,floatX=float32 python convolution_net_sentence.py -nonstatic -word2vec