first pass at turn based conversation

vocode/turn_based/output_device/base_output_device.py

@@ -0,0 +1,9 @@
+from pydub import AudioSegment
+
+
+class BaseOutputDevice:
+    def send_audio(self, audio: AudioSegment) -> None:
+        raise NotImplementedError
+
+    def terminate(self):
+        pass

vocode/turn_based/output_device/speaker_output.py

@@ -0,0 +1,32 @@
+import sounddevice as sd
+import numpy as np
+from pydub import AudioSegment
+
+from vocode.turn_based.output_device.base_output_device import BaseOutputDevice
+
+
+class SpeakerOutput(BaseOutputDevice):
+    DEFAULT_SAMPLING_RATE = 44100
+
+    def __init__(
+        self,
+        device_info: dict,
+        sampling_rate: int = None,
+    ):
+        self.device_info = device_info
+        self.sampling_rate = sampling_rate or int(
+            self.device_info.get("default_samplerate", self.DEFAULT_SAMPLING_RATE)
+        )
+        self.stream = sd.OutputStream(
+            channels=1,
+            samplerate=self.sampling_rate,
+            dtype=np.int16,
+            device=int(self.device_info["index"]),
+        )
+        self.stream.start()
+
+    def send_audio(self, audio_segment: AudioSegment):
+        self.stream.write(np.frombuffer(audio_segment.raw_data, dtype=np.int16))
+
+    def terminate(self):
+        self.stream.close()