is the created at from session history accurately maps to th LiveKit Community #ask-ai

Join Slack

is the created_at from session history accurately ...

# ask-ai

purple-barista-60143

09/10/2025, 1:17 PM

is the created_at from session history accurately maps to the audio?

tall-airline-16756

09/10/2025, 1:17 PM

Thanks for reaching out, harish. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

tall-airline-16756

09/10/2025, 1:18 PM

Thank you for your question! The

created_at

timestamp in session history shows the exact moment when a room was created in LiveKit. LiveKit sets this automatically right when the room is made, so it marks the start of the session. This timestamp is in UTC and is very precise, down to nanoseconds, which helps keep everything consistent. Just a little heads-up: the

created_at

timestamp is specifically about the room creation time and doesn’t directly track audio or participant activity. If you want to connect timestamps to particular audio events or participant actions, it’s a good idea to look at other data points like

joined_at

for participants or check event logs related to audio activities. I hope this helps clarify things for you! Sources: Analytics API | LiveKit Docs | livekit/livekit_analytics.pb.go | livekit-protocol/livekit/protocol/analytics.py

purple-barista-60143

09/10/2025, 1:20 PM

what about created_at field from chat item? does this map to the beginning of the users speech?

tall-airline-16756

09/10/2025, 1:21 PM

Thank you for your question! The

created_at

field in a chat item shows the exact time when that item was created and added to the chat context. It marks the moment the chat message or related event was saved in the chat history, rather than the very start of a user's speech. If you're hoping to match timestamps to the beginning of a user's speech, you might want to explore transcription-related data. For instance, transcription events come with their own timestamps, and transcription segments include

start_time

and

end_time

fields that align with the actual audio timeline. I hope this helps clarify things for you! Sources: livekit-protocol/livekit/protocol/models.pyi | livekit-plugins/livekit-plugins-aws/livekit/plugins/aws/experimental/realti... | Events and error handling | LiveKit Docs

purple-barista-60143

09/10/2025, 2:05 PM

If you're hoping to match timestamps to the beginning of a user's speech, I want to do this

tall-airline-16756

09/10/2025, 2:06 PM

To help you match timestamps to the beginning of a user's speech, you can use LiveKit's Voice Activity Detection (VAD) system. The VAD sends out a

START_OF_SPEECH

event whenever it detects speech, and this event includes a

timestamp

field that accurately marks the exact moment the speech started. For instance, if you're working with the Python SDK, you can listen for this event using the

RecognitionHooks.on_start_of_speech()

callback, which gives you the

timestamp

in seconds. Similarly, in the JavaScript SDK, the

VADEvent

object contains the

timestamp

field for the same purpose. If you'd like, I’m here to help you with more details or examples to make the implementation smoother. Just let me know! Sources: livekit-agents/livekit/agents/stt/stt.py | agents/src/vad.ts

Open in Slack

Previous Next