Hi guys, Have been scratching my head all day but can't seem to get this working. I am getting TTS from eleven labs and I am trying to publish it from the server to the the room as the server participant using rtc-node. It pushes some audio with a lot of noise then throws below error. Error: an RtcError occured: InvalidState - failed to capture frame at AudioSource.captureFrame (/Users/suroorwijdan/Projects/livekit-interview-poc/node_modules/@swift-school-78240/rtc-node/src/audio_source.ts14213) My current code: // Convert text to speech const audioResult = await this.elevenLabsService.textToSpeech( text, sessionConfig.textToSpeechConfig ); let leftoverBuffer = Buffer.alloc(0); for await (const audioChunk of audioResult.audioStream) { if (audioChunk && audioChunk.length > 0) { // 1. Combine the leftover from the previous chunk with the new chunk. const combinedBuffer = Buffer.concat([leftoverBuffer, audioChunk]); // 2. Calculate how many full 16-bit (2-byte) samples we can process now. const processableLength = combinedBuffer.length - (combinedBuffer.length % 2); if (processableLength === 0) { // Not enough data to form a single sample, so just store it and wait for the next chunk. leftoverBuffer = combinedBuffer; continue; } // 3. Create a slice of the buffer that contains only the complete samples. const chunkToProcess = combinedBuffer.subarray(0, processableLength); // 4. The remainder (0 or 1 byte) is our new leftoverBuffer for the next iteration. leftoverBuffer = combinedBuffer.subarray(processableLength); // 5. Safely create the Int16Array from the processed chunk's buffer. // This is now guaranteed to be safe because chunkToProcess has an even number of bytes. const int16AudioData = new Int16Array(chunkToProcess.buffer); // --- Your LiveKit publishing logic remains the same --- if (int16AudioData.length > 0) { const audioFrame = new AudioFrame( int16AudioData, sessionConfig.audioConfig.sampleRate, sessionConfig.audioConfig.channels, Math.trunc(int16AudioData.byteLength / 2) ); const success = this.liveKitService.captureAudioFrame( sessionConfig.alexIdentity, sessionConfig.roomName, this.alexAudioTrackName, audioFrame ); if (!success) { this.logger.warn(

Failed to publish audio frame for ${sessionConfig.alexIdentity}

); break; } } } }