[BUG] Missing speaker name and ID in audio metadata

[jasper-tng]

Bug Description

When accessing the metadata of the audio stream, the user name is empty and the user ID is -1 regardless of who is speaking in the meeting. This issue only appears in the Python library; the Node.js library correctly reports user names and IDs.

Language/Runtime

Python

SDK Version

0.0.2

Environment Details

• OS and Version: macOS 26.2
• Python version: 3.13.11
• Zoom client version: 6.7.2 (72191)

Steps To Reproduce

1. Create a Zoom app that has access to RTMS data as in the sample setup.
2. Configure the Python environment to contain Zoom Client ID and password as in the sample setup.
3. Run a minimal reproducer based on the provided Python example, see below.
4. Start a meeting with the app enabled and start speaking.

Expected Behavior

The Python app prints the user name and user ID of the meeting participant currently speaking: Speaker: <username> (userId: <userID>).

Actual Behavior

The Python app prints Speaker: (userId: -1).

Code Example

import signal
import sys

import rtms
from dotenv import load_dotenv

load_dotenv()

client = rtms.Client()

# Graceful shutdown handler
def signal_handler(sig, frame):
    print('\nShutting down gracefully...')
    client.leave()
    sys.exit(0)

signal.signal(signal.SIGINT, signal_handler)

# Webhook event handler
@client.on_webhook_event()
def handle_webhook(payload):
    if payload.get('event') == 'meeting.rtms_started':
        rtms_payload = payload.get('payload', {})
        client.join(
            meeting_uuid=rtms_payload.get('meeting_uuid'),
            rtms_stream_id=rtms_payload.get('rtms_stream_id'),
            server_urls=rtms_payload.get('server_urls'),
            signature=rtms_payload.get('signature')
        )

# Callback handlers
@client.onJoinConfirm
def on_join(reason):
    print(f'Joined meeting: {reason}')

@client.onAudioData
def on_audio(data, size, timestamp, metadata):
    print(f'Speaker: {metadata.userName} (userId: {metadata.userId})')

@client.onLeave
def on_leave(reason):
    print(f'Left meeting: {reason}')

if __name__ == '__main__':
    print('Webhook server running on http://localhost:8080')
    import time
    while True:
        # Process queued join requests from webhook thread
        client._process_join_queue()
        # Poll for SDK events
        client._poll_if_needed()
        time.sleep(0.01)

Relevant Log Output

Additional Context

It seems this issue is specific to the Python library; the Node.js version correctly reports user names and IDs.

Verification

• I've searched existing issues to ensure this bug hasn't already been reported
• I've verified this bug still exists in the latest version of the SDK
• I've included all necessary information to reproduce this issue

[MaxMansfield]

@jasper-tng Thanks for reporting this! This is the expected behavior if you are using AUDIO_MIXED_STREAMS. If you change the Audio Params to use the AUDIO_MULTI_STREAM it should give you the metadata that you are looking for.

Here's how to set AUDIO_MULTI_STREAMS in Python:

import rtms

# Option 1: Use the default constructor (AUDIO_MULTI_STREAMS is already the default)
params = rtms.AudioParams()
# dataOpt is already 2 (AUDIO_MULTI_STREAMS)

# Option 2: Explicitly set it
params = rtms.AudioParams()
params.dataOpt = 2  # AUDIO_MULTI_STREAMS

# Option 3: Use the constant
params = rtms.AudioParams()
params.dataOpt = rtms.AUDIO_MULTI_STREAMS

# Then apply to client
client.setAudioParams(params)

Let me know if that helps!

[jasper-tng]

@MaxMansfield Thanks for the explanation! Your proposed usage works; the AUDIO_MULTI_STREAM option does make speaker names are available.

I still believe this is a bug, or at least an inconsistency between library implementations, since the Node.js library does provide speaker names even when using AUDIO_MIXED_STREAMS whereas the Python library does not. I generally find the Node.js library behavior useful and appropriate since Zoom can attribute a current speaker even if multiple participants are unmuted. This is especially convenient if the Zoom app that uses the RTMS library only needs a single input stream but wants to differentiate between speakers at any point in time.

[MaxMansfield]

@jasper-tng This is a discussion I've had with our core C++ SDK engineering team but it doesn't seem that they are able to make this change. My understanding is that this is due to only one stream being sent for the entire meeting when using a mixed stream. Instead, you might want to look into the OnActiveSpeakerEvent.

What I noticed with this is that AUDIO_MULTI_STREAMS should have been the default for this SDK. I pushed some fixes for a similar bug (issue #88) and with that made sure that this is applied as a default when using the SDK.

I'll release that fix soon.