Sybill employs advanced AI algorithms to transcribe meeting conversations accurately and attribute speakers to their respective dialogue. However, in some instances, transcription and attribution issues may arise. This guide addresses common challenges related to speaker attribution errors and ad-hoc transcription issues, providing solutions to help you overcome these obstacles effectively. Let's explore each topic in detail:
Speaker Attribution Errors
Sybill's speaker attribution feature is designed to assign dialogue to the correct participant in a meeting. However, occasional errors may occur. Sybill continuously learns and improves its speaker attribution capabilities. If you encounter persistent attribution errors, provide feedback to our support team at help@sybill.ai. Sharing specific instances and details about the errors will aid us in enhancing the system's accuracy.
Adhoc Transcription Issues
While Sybill strives for accurate and comprehensive transcription, there may be instances where adhoc issues affect the transcription quality. Here are some common adhoc transcription issues and ways to address them:
Background Noise or Disturbances: Background noise, overlapping dialogue, or technical disruptions during meetings can impact transcription accuracy. Minimize background noise and ensure optimal audio conditions for improved transcription results.
Uncommon or Technical Terminology: Sybill relies on a service powered by a large language model to understand and transcribe meeting conversations. However, it may encounter challenges with uncommon or technical terminology. You can add Custom Vocabulary for your industry or organization in Settings > Admin > Custom Vocabulary. Sybill will pay special attention to the words that you add under Custom Vocabulary while transcribing.
Accents and Speech Variations: Different accents, speech patterns, and variations in pronunciation can occasionally affect transcription accuracy. Sybill continually learns from user interactions, including feedback on transcription issues. If you encounter challenges specific to accents or speech variations, please share your experiences with our support team at help@sybill.ai.
Please note that Sybill only supports transcription in English at the moment.
Transcription and attribution are vital components of Sybill's capabilities, and Sybill leverages one of the most accurate speaker attribution and transcription systems in the market.