Choose an API

LansonAI provides two live transcription APIs today, plus Voice Agent which is coming soon. Choose based on your use case.

Decision table

Realtime Voice Context

When to use: scenarios where you need text while the person is still speaking.

• Live captions, real-time display
• Real-time translation (source + translated text together) — coming soon for external sessions
• Voice interaction requiring immediate feedback

Characteristics:

• WebSocket full-duplex connection
• Audio format: PCM16LE / 16kHz / mono
• Server-side VAD for automatic speech segment splitting
• Millisecond-level latency (p50 ~310ms)
• Supports concurrent sessions

Constraints:

• Requires continuous audio stream
• Session and idle timeout limits apply

→ Realtime Overview

Recorded Speech

When to use: you already have a complete audio file and need the transcription result.

• Post-meeting processing
• Batch audio file transcription
• When you need precise timestamps
• When you need structured results (summary, chapters)

Characteristics:

• HTTP async (submit → poll / webhook)
• Submit via audio URL
• Automatic slicing + concurrent transcription + aggregation
• Segment-level timestamps
• Optional LLM review for error correction
• Webhook callback support

Constraints:

• Not real-time; requires processing time
• Audio must be reachable via HTTP(S) URL

→ Recorded Overview

Voice Agent

When to use: building voice agents or phone bots.

• Reception / Voice Agent API
• Development primitives for voice robots

Status: Coming soon

Comparison

Next steps

• Realtime Quickstart
• Transcribe Audio
• Authentication