Add gain boosted sound effect variants for Azure with streaming. (#185)

* Files modified to compensate for Azure_TTS issues.

Every thing that has a pedalboard issue will now have two variations

one _Gain_Boost
One _Normal

Gain Boost will have boosted gain that needs to be tied with the output streaming issue :  Basically the Gain Boost option should automatically turn output streaming to enabled as this option is intended to solve the very quiet issue azure TTS / Wingman Pro cause.  This effectively boosts gain upwards to 3800%.

The normal one should force output streaming to disabled, as if output streaming stays on with the boosted gain.. well if you wear headphones it wont be the most pleasant experience ever.

This impacts Low_Quality_Radio, Medium_Quality_Radio, High_End_Radio only.  As these  have the effect boards on them which seems to be the issue.

Also once this is pushed up, the 'Beep' effect needs to be double checked to make sure its playing the apollo_beep as currently I think its playing the old beep.

* use gain boosted variants for Azure with streaming

* convert to 44khz mono wav

* implement Apollo Beep as separate effect

---------

Co-authored-by: Theandril <michaelsizemore90@gmail.com>
Co-authored-by: Simon Hopstätter <simon.hopstaetter@shipbit.de>

Author: 3 people

api/interface.py

@@ -316,7 +316,10 @@ class WingmanProSettings(BaseModel):
 
 class SoundConfig(BaseModel):
     play_beep: bool
-    """adds a beep/Quindar sound before and after the wingman talks"""
+    """adds a Beep/Quindar sound before and after the wingman talks"""
+
+    play_beep_apollo: bool
+    """adds a Apollo Beep sound before and after the wingman talks"""
 
     effects: list[SoundEffect]
     """You can put as many sound effects here as you want. They stack and are added in the defined order here."""

audio_samples/Apollo_Beep.wav

@@ -52,15 +52,22 @@ async def play_audio(
 
         def notify_playback_finished():
             audio_player.playback_events.unsubscribe("finished", playback_finished)
+
             if sound_config.play_beep:
-                audio_player.play_beep()
+                audio_player.play_wav("beep.wav")
+            elif sound_config.play_beep_apollo:
+                audio_player.play_wav("Apollo_Beep.wav")
+
             WebSocketUser.ensure_async(
                 audio_player.notify_playback_finished(wingman_name)
             )
 
         def notify_playback_started():
             if sound_config.play_beep:
-                audio_player.play_beep()
+                audio_player.play_wav("beep.wav")
+            elif sound_config.play_beep_apollo:
+                audio_player.play_wav("Apollo_Beep.wav")
+
             WebSocketUser.ensure_async(
                 audio_player.notify_playback_started(wingman_name)
             )

providers/elevenlabs.py

@@ -285,6 +285,7 @@ def buffer_callback(audio_buffer):
                     buffer_callback,
                     sound_config,
                     wingman_name=wingman_name,
+                    use_gain_boost=True,  # "Azure Streaming" low gain workaround
                 )
             else:
                 await audio_player.play_with_effects(

providers/open_ai.py

@@ -178,6 +178,7 @@ def buffer_callback(audio_buffer):
                 buffer_callback=buffer_callback,
                 config=sound_config,
                 wingman_name=wingman_name,
+                use_gain_boost=True,  # "Azure Streaming" low gain workaround
             )
         else:  # non-streaming
             response = requests.post(

providers/wingman_pro.py

@@ -91,7 +91,9 @@ async def play_with_effects(
             audio = sound_effect(audio, sample_rate)
 
         if config.play_beep:
-            audio = self._add_beep_effect(audio, sample_rate)
+            audio = self._add_wav_effect(audio, sample_rate, "beep.wav")
+        elif config.play_beep_apollo:
+            audio = self._add_wav_effect(audio, sample_rate, "Apollo_Beep.wav")
 
         channels = audio.shape[1] if audio.ndim > 1 else 1
 
@@ -127,10 +129,10 @@ async def notify_playback_finished(self, wingman_name: str):
         if callable(self.on_playback_finished):
             await self.on_playback_finished(wingman_name)
 
-    def play_beep(self):
+    def play_wav(self, audio_sample_file: str):
         bundle_dir = path.abspath(path.dirname(__file__))
         beep_audio, beep_sample_rate = self.get_audio_from_file(
-            path.join(bundle_dir, "../audio_samples/beep.wav")
+            path.join(bundle_dir, f"../audio_samples/{audio_sample_file}")
         )
         self.start_playback(beep_audio, beep_sample_rate, 1, None)
 
@@ -142,10 +144,12 @@ def _get_audio_from_stream(self, stream: bytes) -> tuple:
         audio, sample_rate = sf.read(io.BytesIO(stream), dtype="float32")
         return audio, sample_rate
 
-    def _add_beep_effect(self, audio: np.ndarray, sample_rate: int) -> np.ndarray:
+    def _add_wav_effect(
+        self, audio: np.ndarray, sample_rate: int, audio_sample_file: str
+    ) -> np.ndarray:
         bundle_dir = path.abspath(path.dirname(__file__))
         beep_audio, beep_sample_rate = self.get_audio_from_file(
-            path.join(bundle_dir, "../audio_samples/beep.wav")
+            path.join(bundle_dir, f"../audio_samples/{audio_sample_file}")
         )
 
         # Resample the beep sound if necessary to match the sample rate of 'audio'
@@ -179,6 +183,7 @@ async def stream_with_effects(
         sample_rate=16000,
         channels=1,
         dtype="int16",
+        use_gain_boost=False,
     ):
         buffer = bytearray()
         stream_finished = False
@@ -206,10 +211,15 @@ def callback(outdata, frames, time, status):
             await self.notify_playback_started(wingman_name)
 
             if config.play_beep:
-                self.play_beep()
+                self.play_wav("beep.wav")
+            elif config.play_beep_apollo:
+                self.play_wav("Apollo_Beep.wav")
+
             self.raw_stream.start()
 
-            sound_effects = get_sound_effects(config)
+            sound_effects = get_sound_effects(
+                config=config, use_gain_boost=use_gain_boost
+            )
             audio_buffer = bytearray(buffer_size)
             filled_size = buffer_callback(audio_buffer)
             while filled_size > 0:
@@ -234,6 +244,9 @@ def callback(outdata, frames, time, status):
                 sd.sleep(100)
 
             if config.play_beep:
-                self.play_beep()
+                self.play_wav("beep.wav")
+            elif config.play_beep_apollo:
+                self.play_wav("Apollo_Beep.wav")
+
             self.is_playing = False
             await self.notify_playback_finished(wingman_name)

services/audio_player.py

@@ -10,7 +10,6 @@
     Gain,
     Bitcrush,
     Compressor,
-    PitchShift,
     Distortion,
 )
 from api.interface import SoundConfig
@@ -45,7 +44,7 @@ class SoundEffects(Enum):
             LowpassFilter(cutoff_frequency_hz=3400),
             Resample(target_sample_rate=8000),  # Lower resample rate for tinny effect
             Reverb(room_size=0.1, damping=0.3, wet_level=0.1, dry_level=0.9),
-            Gain(gain_db=-10),
+            Gain(gain_db=-17),
         ]
     )
     MEDIUM_QUALITY_RADIO = Pedalboard(
@@ -56,6 +55,7 @@ class SoundEffects(Enum):
             Resample(target_sample_rate=16000),
             Reverb(room_size=0.01, damping=0.3, wet_level=0.1, dry_level=0.9),
             Compressor(threshold_db=-18, ratio=4),
+            Gain(gain_db=4),
         ]
     )
     HIGH_END_RADIO = Pedalboard(
@@ -65,6 +65,38 @@ class SoundEffects(Enum):
             Compressor(threshold_db=-10, ratio=2),
             Reverb(room_size=0.001, damping=0.3, wet_level=0.1, dry_level=0.9),
             Resample(target_sample_rate=44100),
+            Gain(gain_db=2),
+        ]
+    )
+    LOW_QUALITY_RADIO_GAIN_BOOST = Pedalboard(
+        [
+            Distortion(drive_db=30),
+            HighpassFilter(cutoff_frequency_hz=800),
+            LowpassFilter(cutoff_frequency_hz=3400),
+            Resample(target_sample_rate=8000),  # Lower resample rate for tinny effect
+            Reverb(room_size=0.1, damping=0.3, wet_level=0.1, dry_level=0.9),
+            Gain(gain_db=70),
+        ]
+    )
+    MEDIUM_QUALITY_RADIO_GAIN_BOOST = Pedalboard(
+        [
+            Distortion(drive_db=15),
+            HighpassFilter(cutoff_frequency_hz=300),
+            LowpassFilter(cutoff_frequency_hz=5000),
+            Resample(target_sample_rate=16000),
+            Reverb(room_size=0.01, damping=0.3, wet_level=0.1, dry_level=0.9),
+            Compressor(threshold_db=-18, ratio=4),
+            Gain(gain_db=82),
+        ]
+    )
+    HIGH_END_RADIO_GAIN_BOOST = Pedalboard(
+        [
+            HighpassFilter(cutoff_frequency_hz=100),
+            LowpassFilter(cutoff_frequency_hz=8000),  # Adjust cutoff to avoid conflicts
+            Compressor(threshold_db=-10, ratio=2),
+            Reverb(room_size=0.001, damping=0.3, wet_level=0.1, dry_level=0.9),
+            Resample(target_sample_rate=44100),
+            Gain(gain_db=30),
         ]
     )
 
@@ -103,7 +135,7 @@ class SoundEffects(Enum):
     )
 
 
-def get_sound_effects(config: SoundConfig):
+def get_sound_effects(config: SoundConfig, use_gain_boost: bool = False):
     if config is None or not config.effects or len(config.effects) == 0:
         return []
 
@@ -114,13 +146,18 @@ def get_sound_effects(config: SoundConfig):
         "LOW_QUALITY_RADIO": SoundEffects.LOW_QUALITY_RADIO.value,
         "MEDIUM_QUALITY_RADIO": SoundEffects.MEDIUM_QUALITY_RADIO.value,
         "HIGH_END_RADIO": SoundEffects.HIGH_END_RADIO.value,
+        "LOW_QUALITY_RADIO_GAIN_BOOST": SoundEffects.LOW_QUALITY_RADIO_GAIN_BOOST.value,
+        "MEDIUM_QUALITY_RADIO_GAIN_BOOST": SoundEffects.MEDIUM_QUALITY_RADIO_GAIN_BOOST.value,
+        "HIGH_END_RADIO_GAIN_BOOST": SoundEffects.HIGH_END_RADIO_GAIN_BOOST.value,
         "INTERIOR_SMALL": SoundEffects.INTERIOR_SMALL.value,
         "INTERIOR_MEDIUM": SoundEffects.INTERIOR_MEDIUM.value,
         "INTERIOR_LARGE": SoundEffects.INTERIOR_LARGE.value,
     }
 
     for effect in config.effects:
         effect_name = effect.value
+        if use_gain_boost and f"{effect_name}_GAIN_BOOST" in mapping:
+            effect_name += "_GAIN_BOOST"
         effect = mapping.get(effect_name)
         if effect:
             sound_effects.append(effect)

services/sound_effects.py

@@ -30,7 +30,7 @@ prompts:
 record_key: delete
 sound:
   effects: [AI]
-  play_beep: True
+  play_beep_apollo: true
 openai:
   tts_voice: onyx
 commands:

templates/configs/_Star Citizen/ATC.template.yaml

@@ -47,6 +47,7 @@ features:
 sound:
   effects: []
   play_beep: false
+  play_beep_apollo: false
 openai:
   conversation_model: gpt-4o
   summarize_model: gpt-4o