Transcribe

💡 Definition

Amazon Transcribe makes it easy for developers to add speech-to-text capabilities to their applications. It uses advanced deep learning concepts for automatic speech recognition (ASR).

🔑 Key Concepts

• Speech-to-Text: Converts audio files (MP3, WAV) into written text.
• Speaker Identification: Can identify different speakers in a conversation ("Speaker 1", "Speaker 2").
• Custom Vocabulary: You can add specific terms (e.g., medical jargon, brand names) to improve accuracy.
• Transcribe Medical: specialized version for medical dictation.

⚙️ How it Works

Upload an audio file to S3, start a transcription job. The service processes the audio and outputs a JSON file with the text, timestamps, and confidence scores.

🎯 Use Cases

• Subtitling: Automatically generating captions for videos.
• Call Center Analytics: Transcribing customer support calls for analysis (with Comprehend).
• Meeting Notes: Creating transcripts of meetings.

💰 Pricing Model

• Audio Seconds: Charged per second of audio transcribed.

📝 Exam Tips (CLF-C02)

• Keyword: "Speech to Text", "ASR", "Subtitles".
• Converts Audio -> Text.

---

See Also: * Translate (often used after Transcribe) * Polly (Text to Speech - the opposite)