server playback

audio-service/src/audio_clip.py

@@ -0,0 +1,64 @@
+import scipy.signal
+import scipy.io.wavfile as wavfile
+import numpy as np
+import os
+
+class AudioClip:
+    def __init__(self, metadata, target_sample_rate=44100):
+        """
+        metadata: dict with keys 'filename', 'start', 'end' (seconds)
+        target_sample_rate: sample rate for playback
+        """
+        self.metadata = metadata
+        self.file_path = metadata['filename']
+        self.start = metadata.get('startTime', 0)
+        self.end = metadata.get('endTime', None)
+        self.target_sample_rate = target_sample_rate
+        self.volume = metadata.get('volume', 1.0)
+        self.finished = False
+        self.audio_data, self.sample_rate = self._load_and_process_audio()
+        print(f"AudioClip created for {self.file_path} with start={self.start}s, end={self.end}s, sample_rate={self.sample_rate}Hz, length={len(self.audio_data)/self.sample_rate:.2f}s")
+        self.position = 0  # sample index for playback
+
+    def _load_and_process_audio(self):
+        # Load audio file
+        sample_rate, data = wavfile.read(self.file_path)
+        # Convert to float32
+        if data.dtype != np.float32:
+            data = data.astype(np.float32) / np.max(np.abs(data))
+        # Convert to mono if needed
+        if len(data.shape) > 1:
+            data = np.mean(data, axis=1)
+        # Resample if needed
+        if sample_rate != self.target_sample_rate:
+            num_samples = int(len(data) * self.target_sample_rate / sample_rate)
+            data = scipy.signal.resample(data, num_samples)
+            sample_rate = self.target_sample_rate
+        # Cache only the clip region
+        start_sample = int(self.start * sample_rate)
+        end_sample = int(self.end * sample_rate) if self.end else len(data)
+        cached = data[start_sample:end_sample]
+        cached *= self.volume  # Apply volume
+        return cached, sample_rate
+
+    def get_samples(self, num_samples):
+        # Return next chunk for playback
+        if self.position >= len(self.audio_data):
+            self.finished = True
+            return np.zeros(num_samples, dtype=np.float32)
+        end_pos = min(self.position + num_samples, len(self.audio_data))
+        chunk = self.audio_data[self.position:end_pos]
+        self.position = end_pos
+        if self.position >= len(self.audio_data):
+            self.finished = True
+        # Pad if chunk is short
+        if len(chunk) < num_samples:
+            chunk = np.pad(chunk, (0, num_samples - len(chunk)), mode='constant')
+        return chunk
+
+    def is_finished(self):
+        return self.finished
+
+    def reset(self):
+        self.position = 0
+        self.finished = False