chore: Update TTS dependencies and remove unused imports

kitt/core/__init__.py

@@ -38,7 +38,7 @@ voices = [
     Voice(
         "Freeman",
         neutral=f"{file_full_path}/audio/freeman/neutral.wav",
-        angry="audio/freeman/angry.wav",
+        angry=f"{file_full_path}/audio/freeman/angry.wav",
         speed=1.1,
     ),
     Voice(

kitt/core/model.py

@@ -267,7 +267,7 @@ def run_inference_step(
     #     prompt += "\nThis is the first turn and you don't have <tool_results> to analyze yet."
     prompt += AI_PREAMBLE
 
-    logger.info(f"Prompt is:\n{prompt}")
+    # logger.info(f"Prompt is:\n{prompt}")
 
     if backend == "ollama":
         output = run_inference_ollama(prompt)
@@ -348,7 +348,7 @@ def run_inference_ollama(prompt):
 def run_inference(prompt, backend="ollama"):
     prompt += AI_PREAMBLE
 
-    logger.info(f"Prompt is:\n{prompt}")
+    # logger.info(f"Prompt is:\n{prompt}")
 
     if backend == "ollama":
         output = run_inference_ollama(prompt)

kitt/core/tts.py

@@ -1,3 +1,4 @@
+import copy
 from collections import namedtuple
 
 import soundfile as sf
@@ -54,6 +55,15 @@ voices_replicate = [
 ]
 
 
+def prep_for_tts(text: str):
+    text_tts = copy.deepcopy(text)
+    text_tts = text_tts.replace("km/h", "kilometers per hour")
+    text_tts = text_tts.replace("°C", "degrees Celsius")
+    text_tts = text_tts.replace("°F", "degrees Fahrenheit")
+    text_tts = text_tts.replace("km", "kilometers")
+    return text_tts
+
+
 def voice_from_text(voice, voices):
     for v in voices:
         if voice == f"{v.name} - Neutral":

kitt/core/utils.py

@@ -15,7 +15,7 @@ def plot_route(points, vehicle: Union[tuple[float, float], None] = None):
     # fig = px.line_geo(lat=lats, lon=lons)
     # fig.update_geos(fitbounds="locations")
 
-    fig = px.line_mapbox(lat=lats, lon=lons, color_discrete_sequence=["red"])
+    fig = px.line_mapbox(lat=lats, lon=lons, color_discrete_sequence=["red"], zoom=6)
 
     if vehicle:
         fig.add_trace(

kitt/keepalive.py

@@ -36,7 +36,7 @@ def job():
 
 logger.info("First run to boot up.")
 run_replicate_model()
-schedule.every(100).seconds.do(job)
+schedule.every(90).seconds.do(job)
 logger.info("Keepalive started.")
 
 while True:

kitt/skills/weather.py

@@ -67,8 +67,8 @@ def get_weather(location: str = "here"):
     # Formulate the sentences - {region}, {country}
     weather_sentences = (
         f"The current weather in {location} is {condition_text} "
-        f"with a temperature of {temperature_c} degrees C"
-        f" that feels like {feelslike_c} degrees C."
+        f"with a temperature of {temperature_c} °C"
+        f" that feels like {feelslike_c} °C."
         if feelslike_c != temperature_c
         else ""
         # f"Humidity is at {humidity}%. "
@@ -132,7 +132,7 @@ def get_forecast(city_name: str = "", when=0, **kwargs):
                 number_str = f"in {when-1} days"
 
             # Generate a sentence for the day's forecast
-            forecast_sentence = f"On {date} ({number_str}) in {city_name}, the weather will be {conditions} with a high of {max_temp_c} C and a low of {min_temp_c} C. There's a {chance_of_rain}% chance of rain. "
+            forecast_sentence = f"On {date} ({number_str}) in {city_name}, the weather will be {conditions} with a high of {max_temp_c} °C and a low of {min_temp_c} °C. There's a {chance_of_rain}% chance of rain. "
 
             # number = number + 1
             # Add the sentence to the result

main.py

@@ -19,7 +19,7 @@ from kitt.core import voice_options
 
 # from kitt.core.model import process_query
 from kitt.core.model import generate_function_call as process_query
-from kitt.core.tts import run_melo_tts, run_tts_fast, run_tts_replicate
+from kitt.core.tts import prep_for_tts, run_melo_tts, run_tts_fast, run_tts_replicate
 from kitt.skills import (
     code_interpreter,
     date_time_info,
@@ -38,7 +38,7 @@ from kitt.skills import vehicle_status as vehicle_status_fn
 from kitt.skills.common import config, vehicle
 from kitt.skills.routing import calculate_route, find_address
 
-ORIGIN = "Mondorf-les-Bains, Luxembourg"
+ORIGIN = "Luxembourg, Luxembourg"
 DESTINATION = "Paris, France"
 DEFAULT_LLM_BACKEND = "replicate"
 ENABLE_HISTORY = True
@@ -176,6 +176,7 @@ def run_generic_model(query):
 
 
 def clear_history():
+    logger.info("Clearing the conversation history...")
     history.clear()
 
 
@@ -227,15 +228,16 @@ def run_llama3_model(query, voice_character, state):
         backend=state["llm_backend"],
     )
     gr.Info(f"Output text: {output_text}\nGenerating voice output...")
+    output_text_tts = prep_for_tts(output_text)
     voice_out = None
     if global_context["tts_enabled"]:
         if "Fast" in voice_character:
-            voice_out = run_melo_tts(output_text, voice_character)
+            voice_out = run_melo_tts(output_text_tts, voice_character)
         elif global_context["tts_backend"] == "replicate":
-            voice_out = run_tts_replicate(output_text, voice_character)
+            voice_out = run_tts_replicate(output_text_tts, voice_character)
         else:
             voice_out = tts_gradio(
-                output_text, voice_character, speaker_embedding_cache
+                output_text_tts, voice_character, speaker_embedding_cache
             )[0]
         #
         # voice_out = run_tts_fast(output_text)[0]