Automatic Language Detection allows participants to speak in their preferred language without manually selecting it before or during a meeting. The system detects the spoken language in real time and routes it accordingly, enabling more natural, multilingual workflows. This also applies to Bots used for connecting with External Meeting platforms like Zoom, Teams, and ON24.
How it works
When enabled, KUDO continuously monitors the audio stream and identifies the spoken language on the floor automatically. This removes the need for participants to pre-select a source language, making it especially useful for dynamic, multilingual discussions.
- Supports up to 10 speaking languages simultaneously
- Supports advanced language segmentation & profanity filter for captions
- Detection adjusts in real time when a participant switches languages
- Only precongfigured languages are detected
- Must be enabled during meeting scheduling
What to expect when switching languages
Language detection is not instantaneous. When a participant switches languages, KUDO requires a brief stabilization period before detection accuracy normalizes.
If a participant joins speaking one language and then switches to another, it may take several seconds and multiple spoken phrases for the system to recognize the new language, stabilize detection accuracy, and normalize latency. This is expected behavior, not a malfunction.
Factors that affect detection speed
- Number of active speaking languages. More languages increases detection time.
- Number of participants speaking simultaneously. Overlapping speakers reduce accuracy.
- Frequency of language switching. Rapid switching temporarily degrades performance.
- Language type. Some languages may experience slightly higher latency due to audio segmentation requirements.
Best practices
To get the best results from Automatic Language Detection, follow these speaking guidelines:
- Speak at a normal, steady pace
- Speak clearly and intelligibly
- Stick to one language per speaking segment
- Avoid switching languages mid-sentence
- Avoid speaking over other participants
Troubleshooting
| Issue | Possible cause | Recommended action |
|---|---|---|
| Automatic Language Detection is not available | The feature was not enabled during meeting scheduling | This setting must be configured before the meeting starts. Schedule a new meeting with Automatic Language Detection enabled |
| Spoken language is not being detected correctly | The participant's spoken language does not match the selected speaking language | Select Automatic from the speaking language options, or manually select the correct language from the menu |
| High latency or delayed output | Too many active speaking languages or multiple participants speaking simultaneously | Reduce the number of active speaking languages, or encourage participants to speak one at a time |
| Detection accuracy degrades during a session | Rapid or frequent language switching disrupts the detection model | Allow a brief pause between language changes; the system typically stabilizes within seconds |
| Meeting interface becomes unresponsive | Browser or audio stream issue following a language switch | Refresh the browser and rejoin the meeting |