This message was deleted.

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

i am using nodejs

The error "cannot encode field livekit.TrackEgressRequest.room_name to JSON" in Node.js usually pops up because of some hiccups in how the protobuf message is being serialized. No worries though! Here are a few friendly tips to help you sort it out: • Field Naming: Double-check that you’re using the right field name. For instance, go with

roomName

(camelCase) instead of

room_name

when you’re setting up the request. • Output Configuration: Make sure the

output

field is set up correctly—as either a file output or a WebSocket URL. Like this:

const output = { case: 'websocketUrl', value: '<wss://your-service.com>' };

• Use EgressClient: Try using the

startTrackEgress

method from the

@livekit/server-sdk

package—it makes things a lot easier. For example:

const egressInfo = await egressClient.startTrackEgress('my-room', output, 'TR_XXXXXXXXXXXX');

• Validate Input: Make sure your

roomName

and

trackId

are valid strings without any special characters sneaking in. If you’re still stuck after this, feel free to share the exact code snippet you’re using to create the TrackEgressRequest. I’d be happy to take a closer look and help you figure out what’s going on! Sources: livekit/livekit_egress.pb.go | packages/livekit-server-sdk/src/AccessToken.test.ts | cmd/lk/egress.go

const trackToEgressMap = new Map()*;* /** * Starts an Egress process for each audio track. */ exports*.*startEgressForTrack = async (roomName, sessionId, roomId, participantId, participantName, trackId) => { try { console*.*log(

[Egress] Starting Track Egress process. Room: ${roomName}, TrackID: ${trackId}

)*;* if (*!*sessionId || *!*trackId || *!*roomId || *!*participantId) { console*.*error("[Egress CRITICAL] Required IDs (sessionId, trackId, roomId, participantId) are MISSING!")*;* return; } if (*!*roomName || typeof roomName !== 'string' || roomName*.*trim() === '') { console*.*error(

[Egress CRITICAL] Invalid or empty roomName received: "${roomName}". Cannot start Egress.

)*;* return; } // IMPORTANT: Update your NGROK URL here (for port 6081) const websocketUrl = `wss://03a288d8b202.ngrok-free.app/?sessionId=${sessionId}&trackId=${trackId}&roomId=${roomId}&participantId=${participantId}&participantName=${encodeURIComponent(participantName)}`*;* console*.*log(

[Egress] WebSocket URL created: ${websocketUrl}

)*;* // --- FINAL FIX IS HERE (Based on LiveKit AI Assistant) --- // The 'output' object requires a special structure for 'oneof' fields in protobuf. const output = { case*:* 'websocketUrl'*,* value*:* websocketUrl*,* }*;* // The startTrackEgress function expects three separate arguments. const egressInfo = await egressClient*.*startTrackEgress( roomName*,* output*,* trackId )*;* // --- END FINAL FIX --- console*.log(`[Egress] Track Egress started successfully. Egress ID: ${egressInfo.*_egressId_}`)*;* // Store the Egress ID so we can stop it later trackToEgressMap*.set(trackId,* egressInfo*.*_egressId_)*;* return egressInfo*;* } catch (error) { console*.error(

[Egress] Error starting Track Egress for track ${trackId}:

,* error)*;* // Don't throw the error back to the webhook, just log it // throw error; } }*;* /** * Stops the running Egress for a track. */ exports*.*stopEgressForTrack = async (trackId) => { const egressId = trackToEgressMap*.*get(trackId)*;* if (egressId) { try { console*.*log(

[Egress] Stopping Egress for Track ID: ${trackId}, Egress ID: ${egressId}

)*;* await egressClient*.*stopEgress(egressId)*;* trackToEgressMap*.*delete(trackId)*;* // Cleanup is important console*.*log(

[Egress] Successfully stopped Egress for track ${trackId}

)*;* } catch (error) { // An error can occur if Egress is already stopped, handle it if (error*.message.*includes('Egress not found')) { console*.*warn(

[Egress] Egress already stopped or not found for ID: ${egressId}

)*;* trackToEgressMap*.*delete(trackId)*;* // Still remove from the map } else { console*.error(

[Egress] Error stopping Egress:

,* error)*;* } } } else { console*.*warn(

[Egress] Stop request received but no active Egress found for trackId: ${trackId}

)*;* } }*;*

Thank you so much for sharing your code! From what I can see, it looks like the issue is resolved by properly structuring the

output

field for the protobuf

TrackEgressRequest

. Here are a few important things to keep in mind to make sure everything runs smoothly: • Output Field: You’ve got the

output

field correctly set up as a oneof field with

case: 'websocketUrl'

and

value: websocketUrl

, which fits perfectly with the protobuf requirements. • Room Name Validation: Your check to make sure

roomName

is a non-empty string is great—it helps avoid passing any invalid data. • WebSocket URL: The WebSocket URL you’re generating dynamically looks properly formatted. • Error Handling: Using a try-catch block to log errors without interrupting the webhook flow is a solid approach. If you’re still running into that JSON encoding error, here are a few things you might want to double-check: 1. Make sure

roomName

and

trackId

are valid strings without any special or control characters. 2. Confirm that your

@livekit/server-sdk

package is up-to-date to avoid any compatibility hiccups. 3. Try logging the

output

object right before you pass it to

startTrackEgress

to verify its structure. If the problem keeps popping up, just let me know—we can dig into some more debugging steps together!

what the latest vesriosn for nodejs : @livekit/server-sdk

The latest version of the @livekit/server-sdk for Node.js is 2.13.2. You can easily update to this version by running:

npm install @livekit/server-sdk@2.13.2

Just a quick heads-up—before you upgrade, it’s a good idea to check out the changelog to see if there are any important changes or cool new features. Let me know if you need a hand with it! Sources: pnpm-lock.yaml | packages/livekit-rtc/CHANGELOG.md

how to check my current version

To find out which version of the @livekit/server-sdk you have installed in your project, you’ve got a couple of easy options: • Just run this command: `npm list @livekit/server-sdk`It’ll show you the version you’ve got installed. • Or, you can peek inside your

package.json

file and look for

@livekit/server-sdk

under

dependencies

or

devDependencies

.

in my package.json this library are avilable : livekit-server-sdk"*:* "^2.12.0"*,@livekit/rtc-node":* "^0.13.11"*,*

Your package.json shows that you’re using livekit-server-sdk version 2.12.0 and @livekit/rtc-node version 0.13.11. If you’d like to update to the latest livekit-server-sdk (which is 2.13.2 right now), just run this command:

npm install @livekit/server-sdk@2.13.2

Feel free to reach out if you want a hand with the update or have any other questions—I’m here to help!

'@livekit/server-sdk@2.13.2' is not in this registry.

The latest version of @livekit/server-sdk is confirmed to be 2.13.2. If you’re running into any trouble with the npm registry, no worries—here are a few simple steps you can try: • Clear your npm cache by running:

npm cache clean --force

• Make sure you’re connected to the right npm registry by running: `npm config get registry`The default should be

<https://registry.npmjs.org/>

. If it’s not, you can set it with:

npm config set registry <https://registry.npmjs.org/>

• Give the installation another shot with:

npm install @livekit/server-sdk@2.13.2

If you’re still having issues or want a hand with anything else, just let me know—I’m here to help! Sources: packages/livekit-server-sdk/CHANGELOG.md | pnpm-lock.yaml

Hello! I have successfully set up a

TrackEgress

to stream a single audio track to my Node.js server via a WebSocket URL. My server is receiving audio chunks from the Egress WebSocket (typically 1920 bytes per message), and I am piping this raw audio data directly to the Google Cloud Speech-to-Text streaming API. The problem is that while Google STT sends back responses, the

transcript

field is always empty, as if it's processing silence. My Google STT configuration is set to

encoding: 'LINEAR16'

,

sampleRateHertz: 16000

, and

audioChannelCount: 1

. My question is: What is the exact audio format of the data sent from a TrackEgress WebSocket stream? Is it guaranteed to be raw PCM

LINEAR16

at 16kHz, mono? Is there any header or metadata within the stream that I need to strip before forwarding the audio chunks to the STT API?

The audio data sent from a TrackEgress WebSocket stream in LiveKit comes in raw PCM LINEAR16 (pcm_s16le) format, with a content type of audio/x-raw. Just a heads-up—the sample rate is usually 48kHz, not 16kHz, and it matches the incoming audio stream's settings. To make sure it works smoothly with the Google Cloud Speech-to-Text API, you might need to resample the audio down to 16kHz mono before sending it over. Also, since there are no headers or metadata in the stream, you can process the audio chunks directly as raw PCM data. If you'd like, I’m here to help you with resampling or anything else you need for your setup! Just let me know. Sources: Universal Egress | Track Egress | LiveKit Docs | Universal Egress | Universal Egress