how i can get access to participant identity in this case

@session.on("user_input_transcribed")
    def on_transcript(ev):

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

ev has no participant

Thanks — good detail. You're correct:

ev

for

user_input_transcribed

does not include a

participant

field. Use the session's linked participant instead. For example:

@session.on("user_input_transcribed")
def on_transcript(ev):
    linked = session.room_io.linked_participant
    if linked:
        print("participant identity:", linked.identity)
    else:
        print("no linked participant (session.room_io.linked_participant is None)")

If you need to target a specific participant in multi‑participant rooms, create a

RoomIO

bound to that identity when you start the session:

room_io = RoomIO(session, room=ctx.room, participant="desired-participant-identity")

If that doesn't match your setup, please share: • the printed structure of

ev

(e.g.,

print(ev.__dict__)

) • the

livekit-agents

package version and I’ll guide the exact fix. Sources: Agent session | LiveKit Documentation | How to set a fixed participant for agent audio processing, with interactive elements

i get only this for env {'type': 'user_input_transcribed', 'transcript': 'hello there everyone this is a demo test of me you saw about that trying', 'is_final': False, 'speaker_id': None, 'language': 'en-US', 'created_at': 1778069614.004799}

Thanks — that event dict confirms it:

user_input_transcribed

does not include a participant object. The transcription payload you showed (

type

,

transcript

,

is_final

,

speaker_id

, etc.) is expected and won't contain the participant identity. What you can do instead: • Use the session's linked participant:

session.room_io.linked_participant

(or

session.linked_participant

depending on your setup). This is the participant the session is currently wired to for audio I/O.

@session.on("user_input_transcribed")
def on_transcript(ev):
    linked = session.room_io.linked_participant
    if linked:
        print("participant identity:", linked.identity)
    else:
        print("no linked participant")

• If you need transcripts tied to a specific participant in a multi-participant room, create the session with a RoomIO bound to that identity:

room_io = RoomIO(session, room=ctx.room, participant="desired-participant-identity")

Note:

speaker_id

in the event may be None in many setups; it is not a reliable substitute for participant identity unless your configuration explicitly populates it. If you want, share your

livekit-agents

version and how you create the session/RoomIO and I’ll suggest the exact code changes. , with interactive elements

i want to get identity using speaker_id

Thanks — I’m responding to your goal: use

speaker_id

to get the participant identity. Short answer: you cannot reliably map

speaker_id

to LiveKit

participant.identity

.

speaker_id

comes from STT speaker diarization (labels like "A", "B") and is not the same as LiveKit participant identity. It may also be None for short utterances. What you can do: • If you only need one participant’s audio, bind the session to that participant so all transcripts belong to that identity:

room_io = RoomIO(session, room=ctx.room, participant="desired-participant-identity")

• For multi‑participant rooms, restrict which participant(s) feed the STT by configuring

RoomInputOptions

/

RoomIO

so you only process known identities. • Enable diarization in your STT config (e.g.,

speaker_labels=True

for AssemblyAI) if you want diarization output — but note diarization still won’t provide

participant.identity

. If you need a manual mapping approach, I can suggest workflows (for example, creating short registration prompts per participant to map diarization labels to identities), but I’ll need: • your

livekit-agents

SDK version, and • which STT provider and diarization settings you’re using. Which of those would you like to try? Sources: AssemblyAI STT | LiveKit Documentation | Deepgram STT | LiveKit Documentation | Speech-to-text (STT) models overview | LiveKit Documentation | Agent session | LiveKit Documentation | How to set a fixed participant for agent audio processing, with interactive elements