Add auto append text for punctuation

mimic3_http/synthesis.py

@@ -14,7 +14,6 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 import argparse
-import asyncio
 import io
 import logging
 import threading
@@ -131,9 +130,7 @@ def do_synthesis_proc(args: argparse.Namespace, request_queue: Queue):
                     _LOGGER.exception("Error during inference")
 
                     # Signal error on main loop
-                    item.loop.call_soon_threadsafe(
-                        item.future.set_exception, e
-                    )
+                    item.loop.call_soon_threadsafe(item.future.set_exception, e)
 
     except Exception:
         _LOGGER.exception("Unexpected error in inference thread")

mimic3_tts/config.py

@@ -262,7 +262,13 @@ class InferenceConfig:
     noise_w: float = 0.8
 
     minor_break_ms: typing.Optional[int] = None
+    """Automatically add milliseconds of silence after a minor break (comma)"""
+
     major_break_ms: typing.Optional[int] = None
+    """Automatically add milliseconds of silence after a major break (period)"""
+
+    auto_append_text: typing.Optional[str] = None
+    """Automatically append text to the end of an utterance if not present (e.g., punctuation)"""
 
 
 @dataclass

mimic3_tts/tts.py

@@ -364,13 +364,19 @@ class Mimic3TextToSpeechSystem(TextToSpeechSystem):
     def begin_utterance(self):
         pass
 
-    # pylint: disable=arguments-differ
     def speak_text(self, text: str, text_language: typing.Optional[str] = None):
         voice = self._get_or_load_voice(self.voice)
 
+        # Automatically append text (e.g., punctuation) if not present
+        append_text = voice.config.inference.auto_append_text
+        if append_text and (not text.endswith(append_text)):
+            text += append_text
+
+        # Automatic silence after major/minor breaks (optional)
         minor_break_ms = voice.config.inference.minor_break_ms
         major_break_ms = voice.config.inference.major_break_ms
 
+        # Process chunks
         for sent_phonemes, break_type in voice.text_to_phonemes(
             text, text_language=text_language
         ):