From 1722b1659af5870f53bf8f03164173abdaea1ffe Mon Sep 17 00:00:00 2001
From: Eren Golge <egolge@mozilla.com>
Date: Wed, 21 Nov 2018 18:01:35 +0100
Subject: [PATCH] phonem updates

---
 datasets/TTSDataset.py |  4 ++--
 utils/text/__init__.py | 33 +++++++++++++++++++++++++++------
 2 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/datasets/TTSDataset.py b/datasets/TTSDataset.py
index 36dde917..c48b9234 100644
--- a/datasets/TTSDataset.py
+++ b/datasets/TTSDataset.py
@@ -6,7 +6,7 @@ import torch
 import random
 from torch.utils.data import Dataset
 
-from utils.text import text_to_sequence, phoneme_to_sequence
+from utils.text import text_to_sequence, phonem_to_sequence
 from utils.data import (prepare_data, pad_per_step, prepare_tensor,
                         prepare_stop_target)
 
@@ -87,7 +87,7 @@ class MyDataset(Dataset):
         else:
             text, wav_file = self.items[idx]
             text = np.asarray(
-                phoneme_to_sequence(text, [self.cleaners]), dtype=np.int32)
+                phonem_to_sequence(text, [self.cleaners]), dtype=np.int32)
             wav = np.asarray(self.load_wav(wav_file), dtype=np.float32)
             sample = {'text': text, 'wav': wav, 'item_idx': self.items[idx][1]}
         return sample
diff --git a/utils/text/__init__.py b/utils/text/__init__.py
index 0ff3110f..bbde1d04 100644
--- a/utils/text/__init__.py
+++ b/utils/text/__init__.py
@@ -3,11 +3,7 @@
 import re
 import phonemizer
 from utils.text import cleaners
-<<<<<<< HEAD
 from utils.text.symbols import symbols, phonemes, _punctuations
-=======
-from utils.text.symbols import symbols, phonemes
->>>>>>> phonem extraction for training
 
 # Mappings from symbol to numeric ID and vice versa:
 _symbol_to_id = {s: i for i, s in enumerate(symbols)}
@@ -71,9 +67,18 @@ def sequence_to_phoneme(sequence):
     return result.replace('}{', ' ')
 
 
-def phoneme_to_sequence(text, cleaner_names):
+def phonem_to_sequence(text, cleaner_names):
+    '''
+    TODO: This ignores punctuations
+    '''
     sequence = []
-    sequence += _phonem_to_sequence(_clean_text(text, cleaner_names))
+    clean_text = _clean_text(text, cleaner_names)
+    for word in clean_text.split():
+        phonems_text = text2phone(word)
+        if phonems_text == None:
+            continue
+        sequence += _phonem_to_sequence(phonems_text)
+        sequence.append(_phonemes_to_id[' '])
     sequence.append(_phonemes_to_id['~'])
     return sequence
 
@@ -122,6 +127,17 @@ def sequence_to_text(sequence):
     return result.replace('}{', ' ')
 
 
+def sequence_to_phonem(sequence):
+    '''Converts a sequence of IDs back to a string'''
+    result = ''
+    for symbol_id in sequence:
+        if symbol_id in _id_to_phonemes:
+            s = _id_to_phonemes[symbol_id]
+            print(s)
+            result += s
+    return result.replace('}{', ' ')
+
+
 def _clean_text(text, cleaner_names):
     for name in cleaner_names:
         cleaner = getattr(cleaners, name)
@@ -135,8 +151,13 @@ def _symbols_to_sequence(symbols):
     return [_symbol_to_id[s] for s in symbols if _should_keep_symbol(s)]
 
 
+<<<<<<< HEAD
 def _phoneme_to_sequence(phonemes):
     return [_phonemes_to_id[s] for s in list(phonemes) if _should_keep_phoneme(s)]
+=======
+def _phonem_to_sequence(phonemes):
+    return [_phonemes_to_id[s] for s in phonemes.split(" ") if _should_keep_phonem(s)]
+>>>>>>> phonem updates
 
 
 def _arpabet_to_sequence(text):