update for malyalam

README.md

@@ -156,6 +156,7 @@ The following environment variables can be configured:
 |----------|---------|-------------|
 | `API_KEY` | hackathon-2026-voice-auth-api-key | API authentication key |
 | `MAX_AUDIO_SIZE_MB` | 10 | Maximum audio file size |
+| `MIN_AUDIO_DURATION_SECONDS` | 5 | Minimum audio duration |
 | `MAX_AUDIO_DURATION_SECONDS` | 60 | Maximum audio duration |
 | `HF_ANTISPOOF_MODEL` | garystafford/wav2vec2-deepfake-voice-detector | HF model name for anti-spoofing |
 | `HF_ANTISPOOF_THRESHOLD` | 0.9 | AI detection threshold |
@@ -193,8 +194,9 @@ Common HTTP status codes:
 
 ## Limitations
 
-- Maximum audio file size: 10MB
-- Maximum audio duration: 60 seconds
+- Maximum audio file size: 10MB
+- Minimum audio duration: 5 seconds
+- Maximum audio duration: 60 seconds
 - Audio format: MP3 only
 - Supported languages: Tamil, English, Hindi, Malayalam, Telugu
 

app/audio_processor.py

@@ -23,10 +23,11 @@ class AudioProcessingError(Exception):
 class AudioProcessor:
     """Process audio from Base64 to normalized waveform."""
     
-    def __init__(self):
-        self.target_sample_rate = settings.TARGET_SAMPLE_RATE
-        self.max_size_mb = settings.MAX_AUDIO_SIZE_MB
-        self.max_duration_seconds = settings.MAX_AUDIO_DURATION_SECONDS
+    def __init__(self):
+        self.target_sample_rate = settings.TARGET_SAMPLE_RATE
+        self.max_size_mb = settings.MAX_AUDIO_SIZE_MB
+        self.max_duration_seconds = settings.MAX_AUDIO_DURATION_SECONDS
+        self.min_duration_seconds = settings.MIN_AUDIO_DURATION_SECONDS
     
     def decode_base64(self, base64_content: str) -> bytes:
         """
@@ -92,11 +93,15 @@ class AudioProcessor:
             if len(audio_array.shape) > 1:
                 audio_array = np.mean(audio_array, axis=1)
             
-            duration = len(audio_array) / sample_rate
-            if duration > self.max_duration_seconds:
-                raise AudioProcessingError(
-                    f"Audio too long: {duration:.1f}s (max {self.max_duration_seconds}s)"
-                )
+            duration = len(audio_array) / sample_rate
+            if duration < self.min_duration_seconds:
+                raise AudioProcessingError(
+                    f"Audio too short: {duration:.2f}s (min {self.min_duration_seconds}s)"
+                )
+            if duration > self.max_duration_seconds:
+                raise AudioProcessingError(
+                    f"Audio too long: {duration:.1f}s (max {self.max_duration_seconds}s)"
+                )
             
             logger.info(f"Loaded audio: {duration:.2f}s @ {sample_rate}Hz")
             return audio_array, sample_rate

app/config.py

@@ -18,6 +18,7 @@ class Settings(BaseSettings):
     # Audio Processing
     MAX_AUDIO_SIZE_MB: int = 10
     MAX_AUDIO_DURATION_SECONDS: int = 60
+    MIN_AUDIO_DURATION_SECONDS: int = 5
     TARGET_SAMPLE_RATE: int = 16000
     
     # Model Configuration

plans/api-specification.md

@@ -182,6 +182,7 @@ The API generates explanations based on detected acoustic features:
 
 - **Requests per minute**: 60
 - **Max audio size**: 10MB (Base64 decoded)
+- **Min audio duration**: 5 seconds
 - **Max audio duration**: 60 seconds
 
 ## HTTP Status Codes