fix(kokoro): trim trailing silence and run-on noise on short prompt synthesis (#960)

backend/backends/__init__.py

@@ -369,6 +369,7 @@ def _get_non_qwen_tts_configs() -> list[ModelConfig]:
             engine="kokoro",
             hf_repo_id="hexgrad/Kokoro-82M",
             size_mb=350,
+            needs_trim=True,
             languages=["en", "es", "fr", "hi", "it", "pt", "ja", "zh"],
         ),
     ]

backend/backends/kokoro_backend.py

@@ -288,7 +288,10 @@ class KokoroTTSBackend:
                 # Return 1 second of silence as fallback
                 return np.zeros(KOKORO_SAMPLE_RATE, dtype=np.float32), KOKORO_SAMPLE_RATE
 
-            audio = np.concatenate(audio_chunks)
-            return audio.astype(np.float32), KOKORO_SAMPLE_RATE
+            audio = np.concatenate(audio_chunks).astype(np.float32)
+            from ..utils.audio import trim_tts_output
+
+            audio = trim_tts_output(audio, sample_rate=KOKORO_SAMPLE_RATE)
+            return audio, KOKORO_SAMPLE_RATE
 
         return await asyncio.to_thread(_generate_sync)