Speech Translation Implementation Guide

Client Side (index.html)

The main HTML file contains three core components:

1. Audio Capture & Processing

• Handles microphone input
• Processes audio chunks
• Converts to WAV format
• Sample rate: 16000Hz
• Buffer size: 2048

2. WebSocket Communication

• Manages real-time streaming
• Handles chunked file uploads
• Processes server responses
• Automatic reconnection

3. File Upload System

• Chunked file processing
• Progress tracking
• Base64 conversion

Server Side (FastAPI)

WebSocket Endpoints

1. Real-time Speech (/ws) Input: { "type": "audio_chunk", "data": "data:audio/wav;base64,...", "language": "en" }

Output: { "type": "transcription", "text": "Transcribed text..." }

2. File Upload (/ws/transcribe) Input: { "type": "file_chunk", "data": "data:audio/...;base64,...", "chunk_number": 1, "total_chunks": 10, "language": "en" }

Responses:

• Chunk Receipt: { "type": "chunk_received", "chunk": 1 }

• Segment Update: { "type": "segment", "number": 1, "text": "Segment text...", "start": 0.0, "end": 2.5 }

• Complete: { "type": "complete", "text": "Full transcription..." }

HTTP Endpoint

POST /transcribe

• Input: multipart/form-data (audio file)
• Output: {"text": "Transcribed text..."}
• Error: {"success": false, "error": "Description"}

Quick Setup

1. Server:

docker build -t whisper-api .
docker run -p 8000:8000 whisper-api