add tts

vibe_bot/tts.py

@@ -0,0 +1,53 @@
+import numpy as np
+import soundfile as sf
+from io import BytesIO
+import os
+import logging
+from kokoro_tts import Kokoro, chunk_text, process_chunk_sequential
+
+logger = logging.getLogger(__name__)
+
+# Default voice settings
+DEFAULT_VOICE = "af_sarah"
+DEFAULT_SPEED = 1.0
+DEFAULT_LANG = "en-us"
+
+
+class TTSEngine:
+    def __init__(self, model_path: str, voices_path: str):
+        self.model_path = model_path
+        self.voices_path = voices_path
+        self.kokoro = Kokoro(model_path, voices_path)
+        logger.info("Kokoro TTS engine initialized")
+
+    def generate_audio(self, text: str, voice: str = DEFAULT_VOICE, speed: float = DEFAULT_SPEED, lang: str = DEFAULT_LANG) -> BytesIO:
+        """Convert text to audio and return as BytesIO (MP3 format)."""
+        all_samples = []
+        sample_rate = None
+
+        chunks = chunk_text(text)
+        logger.info(f"Split text into {len(chunks)} chunks")
+
+        for i, chunk in enumerate(chunks):
+            try:
+                samples, sr = process_chunk_sequential(chunk, self.kokoro, voice, speed, lang)
+                if samples is not None:
+                    if sample_rate is None:
+                        sample_rate = sr
+                    all_samples.append(samples)
+                    logger.info(f"Processed chunk {i+1}/{len(chunks)}")
+            except Exception as e:
+                logger.error(f"Error processing chunk {i+1}: {e}")
+                continue
+
+        if not all_samples:
+            raise ValueError("No audio samples generated - text may be invalid or too long")
+
+        combined = np.concatenate(all_samples)
+
+        buffer = BytesIO()
+        sf.write(buffer, combined, sample_rate, format="MP3", subtype="MPEG_LAYER_III")
+        buffer.seek(0)
+
+        logger.info(f"Generated MP3 audio: {len(combined)} samples at {sample_rate}Hz")
+        return buffer