Feature/tts

afkode/ios/speech.py

@@ -6,13 +6,15 @@
 import os
 import sys
 import time
+from pathlib import Path
 
+import sound
 import speech
 
-# import sound
+from afkode import api, utils
 
 
-def speak(text: str) -> None:
+def speak_old(text: str) -> None:
     """Say the text using iOS capability.
 
     Args:
@@ -25,6 +27,33 @@ def speak(text: str) -> None:
         time.sleep(0.1)
 
 
+def speak(text: str) -> None:
+    """Run text-to-speech on ios platform.
+
+    Args:
+        text: full text to be spoken.
+    """
+    tts_base_path = Path(utils.get_base_path(), "data", "text_to_speech")
+    if not tts_base_path.exists():
+        tts_base_path.mkdir(parents=True)
+    logging.info(f">>>{text}")
+
+    # Attempt to use cache if couple of words or is a command
+    if len(text.split(" ")) <= 2 or text in utils.get_spoken_command_list():
+        tts_path = Path(tts_base_path, f"{text}.wav")
+        if not tts_path.exists():
+            api.google_tts(tts_path, text_input=text)
+    else:
+        tts_path = Path(tts_base_path, "latest.wav")
+        api.google_tts(tts_path, text_input=text)
+
+    player = sound.Player(str(tts_path))
+
+    player.play()
+    while player.playing:
+        time.sleep(0.1)
+
+
 def play_blip() -> None:
     """Play a short sound to help notify of API speed."""
     # sound.play_effect("ui:mouseclick1")
@@ -34,7 +63,6 @@ def play_blip() -> None:
 if __name__ == "__main__":
     # Basic speech testing. Can't run as module, so we import our library here
     sys.path.append(os.path.realpath("../.."))
-    from afkode import utils
 
     utils.setup_logging(log_level=logging.INFO)
     speak("This is a test of the speech module")

afkode/macos/speech.py

@@ -14,11 +14,11 @@
 from pydub import AudioSegment
 from pydub.playback import play
 
-from afkode import utils
+from afkode import api, utils
 
 
 def text_to_speech(sentence: str, idx: int) -> str:
-    """Create a gTTS object.
+    """Create a Google text to speech object.
 
     Args:
         sentence: input text to be spoken
@@ -91,7 +91,7 @@ def split_text(text: str) -> List[str]:
     return splits
 
 
-def speak(text: str) -> None:
+def speak_old(text: str) -> None:
     """Run text-to-speech on non-ios platform.
 
     Args:
@@ -117,6 +117,29 @@ def speak(text: str) -> None:
     q.join()
 
 
+def speak(text: str) -> None:
+    """Run text-to-speech on non-ios platform.
+
+    Args:
+        text: full text to be spoken.
+    """
+    tts_base_path = Path(utils.get_base_path(), "data", "text_to_speech")
+    if not tts_base_path.exists():
+        tts_base_path.mkdir(parents=True)
+    logging.info(f">>>{text}")
+
+    # Attempt to use cache if couple of words or is a command
+    if len(text.split(" ")) <= 2 or text in utils.get_spoken_command_list():
+        tts_path = Path(tts_base_path, f"{text}.wav")
+        if not tts_path.exists():
+            api.google_tts(tts_path, text_input=text)
+    else:
+        tts_path = Path(tts_base_path, "latest.wav")
+        api.google_tts(tts_path, text_input=text)
+
+    play(AudioSegment.from_file(tts_path))
+
+
 def play_blip() -> None:
     """Play a short sound to help notify of API speed."""
     file_name = Path("afkode", "macos", "blip.wav")

afkode/prompts/goal/dnd_partial_summary.txt

@@ -0,0 +1,5 @@
+Your task is to create a summary of the following part of a dungeons and dragons session.
+
+The setting for this one-shot Dungeons and Dragons campaign is a 1990s small-town aesthetic called Pin Tweaks. For context, there are three players who have recently taken a break from the playing session and want a recap of the interesting events and character interactions as they get back into the game.
+
+You should mention what each of the characters has done (and which NPC characters they have engaged with.

afkode/prompts/output/limit_400.txt

@@ -0,0 +1 @@
+Limit your total response to 400 words or less.

afkode/prompts/output/ssml.txt

@@ -0,0 +1,4 @@
+Your goal is to transform the following text into valid SSML Speech Synthesis Markup Language.
+The aim is to prepare the text to be spoken aloud using text to voice.
+Make the text as interesting as possible by using the prosody tag functions where it makes sense.
+----