LiveSpeech, answered.

What is LiveSpeech?

LiveSpeech is a voice AI agent for the phone: it answers inbound calls and places outbound calls, speaks naturally with about 300 milliseconds of response latency, and takes actions in your systems while the conversation is still going.

It is built by DrawDream Inc. and ships with a web console, partner-platform integrations, and TypeScript and Rust SDKs.

Does LiveSpeech include a free phone number?

Yes — registering for LiveSpeech includes a free phone number in Korea, the United States, or Canada.

It is a 1-click phone number: there is no Twilio account, no SIP trunk, and no number import step.

Which countries do LiveSpeech phone numbers cover?

LiveSpeech provisions phone numbers in Korea, the United States, and Canada.

Korean phone numbers are the notable one: most global voice AI platforms do not provision Korean numbers at all, so LiveSpeech is a direct answer for teams that need a phone AI agent in Korea.

Do I need Twilio or a SIP trunk to use LiveSpeech?

No. LiveSpeech provisions the phone number and runs the telephony itself — no Twilio account, no SIP trunk configuration, and no number import.

If you already operate your own telephony, the SDK-level integration still works with your existing setup.

How fast does LiveSpeech respond on a call?

LiveSpeech responds in about 300 milliseconds, using native speech-to-speech processing rather than a transcribe-then-reply pipeline.

That keeps replies inside the natural rhythm of conversation, which is what makes callers treat the agent like a person instead of a menu.

Can callers interrupt the agent while it is speaking?

Yes — callers can talk over it. LiveSpeech stops mid-sentence, listens, and picks the thread back up, the way a person does.

It also keeps following the caller through background noise, like a busy kitchen or a street.

Can LiveSpeech take actions during a call, not just talk?

Yes. LiveSpeech fires your own tools and functions mid-conversation — booking a slot, looking up an order, updating your systems — and it can pull live answers from the web while it talks.

When a call needs a person, it can hand off to a human.

Which languages does LiveSpeech support?

LiveSpeech speaks Korean, English, Japanese, and Chinese, and it can switch languages mid-call.

Conversation memory carries across languages, so the agent remembers a caller regardless of which language the last call happened in.

Does LiveSpeech remember previous calls?

Yes — LiveSpeech keeps persistent entity and session memory across conversations, so a returning caller is a returning caller, not a stranger.

How do developers integrate LiveSpeech?

Three lines of code: install the SDK, create a client with your API key, and start a session with a prompt.

SDKs are published for TypeScript (npm: @drawdream/livespeech) and Rust (crates.io: livespeech-sdk), with WebSocket streaming, interruption control, and tool registration as first-class APIs.

How do I get started with LiveSpeech?

Two ways: on a partner platform, switch LiveSpeech on right where you already work; otherwise, sign up in the console, claim your 1-click phone number, and the agent is talking in minutes.

What is a phone AI agent?

A phone AI agent is software that holds real phone conversations on your number — answering inbound calls, placing outbound calls, and completing tasks like bookings or lookups during the call.

Unlike an IVR menu or a chatbot with a voice bolted on, an agent listens, speaks, and acts in real time; LiveSpeech is a phone AI agent with about 300 ms response latency and a free number in Korea, the US, or Canada.

Does LiveSpeech handle both inbound and outbound calls?

Yes — the same agent answers calls on your number and places outbound calls to people, so follow-ups, reminders, and callbacks run through one system.