INNER CODE UNIT · Python

crossfade_samples

devnen/Kitten-TTS-Server · server.py:601

            crossfade_samples = int(0.01 * engine_output_sample_rate)  # 10ms crossfade

            merged_audio = []
            for i, chunk in enumerate(all_audio_segments_np):
                if i == 0:
                    merged_audio.append(chunk)
                else:
                    # Add silence gap between chunks
                    merged_audio.append(silence_array)

                    # Then add the next chunk with optional crossfade
                    if (
                        len(merged_audio[-2]) >= crossfade_samples
                        and len(chunk) >= crossfade_samples
                    ):
                        # Apply fade out to end of previous audio (before silence)
                        fade_out = np.linspace(1, 0, crossfade_samples)
                        merged_audio[-2][-crossfade_samples:] *= fade_out

View source record →