Update README.md

README.md

@@ -10,18 +10,19 @@ health_check_path: /health
 
 # Talklas API
 
-This FastAPI app is deployed on Hugging Face Spaces for audio transcription, translation, and text-to-speech (TTS). The model loads all at once and includes a feature for detecting inappropriate language. It includes the following endpoints:
+This FastAPI app is deployed on Hugging Face Spaces for audio transcription, translation, and text-to-speech (TTS). The application loads all necessary models at startup and updates each model on demand, including a feature for detecting inappropriate language. It includes the following endpoints:
 
 - `/`: Returns a simple health check response.
 - `/health`: Health check endpoint for Hugging Face Spaces.
 - `/update-languages`: Updates the source and target languages for STT and TTS models.
 - `/translate-text`: Translates text and converts it to speech.
 - `/translate-audio`: Transcribes audio, translates the text, and converts the translated text to speech. Includes speech detection to handle silent audio gracefully.
+- `/text-to-speech`: This will generate a speech from the text and target language given to the server
 
 ## Features
 
 - **Speech Detection**: The `/translate-audio` endpoint detects if the audio is silent (no speech) and returns a user-friendly response.
-- **Transcription (STT)**: Uses MMS or Whisper models to transcribe audio.
+- **Transcription (STT)**: Uses Whisper for English and Tagalog source language and MMS for other Philippine languages to transcribe audio.
 - **Translation (MT)**: Uses the NLLB-200 model to translate text between supported languages.
 - **Text-to-Speech (TTS)**: Uses MMS-TTS models to convert translated text to speech.