[add] Load voice parameters from .env file

Configure Piper TTS synthesis via environment variables (speaker_id,
length_scale, noise_scale, noise_w_scale, volume) loaded from .env
with python-dotenv. Includes .env.example as reference template.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.env.example

@@ -0,0 +1,10 @@
+# Piper TTS voice configuration
+VOICE_MODEL=en_GB-alan-medium.onnx       # Path to Piper .onnx voice model (relative to project root)
+VOICE_SPEAKER_ID=                         # Speaker ID for multi-speaker models (integer, leave empty for default)
+VOICE_LENGTH_SCALE=                       # Speech speed: >1.0 = slower, <1.0 = faster (leave empty for model default)
+VOICE_NOISE_SCALE=                        # Phoneme variability: 0.0 = robotic, 1.0 = expressive (leave empty for model default)
+VOICE_NOISE_W_SCALE=                      # Phoneme width variability (leave empty for model default)
+VOICE_VOLUME=1.0                          # Playback volume multiplier (default: 1.0)
+
+# Vosk STT configuration
+STT_MODEL=vosk-model-small-en-us-0.15    # Vosk model directory name (relative to project root)