According to AssemblyAI, AssemblyAI announced that it has significantly improved its speaker segmentation service, improving speaker identification accuracy by up to 13% compared to previous models. With this update, AssemblyAI stays at the forefront of the market, delivering the best performance for identifying accurate speaker counts among a variety of providers.
Speaker split upgrade
With the latest upgrade, AssemblyAI’s Speaker Diarization now supports 16 languages, including the newly added Chinese, Hindi, Japanese, Korean, and Vietnamese. The goal of these improvements is to improve the readability of transcripts and enable features such as automated video clipping, call coaching, and automated dubbing.
Speaker segmentation is critical to accurately identifying and separating different speakers in audio recordings, which is essential for a variety of applications including meeting transcription, media production, and customer service automation.
New billing notification feature
In addition to Speaker Diarization improvements, AssemblyAI has introduced a new billing notification feature. This tool allows users to set balance thresholds in their billing dashboard and receive email notifications when their balance reaches the specified limit. This proactive approach helps users effectively manage their spending and avoid unexpected costs.
Additional updates
AssemblyAI has also extended the Universal-1 model to support German, enabling accurate and fast transcription in English, Spanish, and German. The company has also enhanced its Personally Identifiable Information (PII) text redaction and item detection models to increase accuracy in detecting and removing sensitive information from transcripts.
Community and Resources
AssemblyAI encourages users to share their projects and join the conversation on social media platforms using the hashtag #AssemblyAI. The company has also provided a variety of resources and guides to help users implement Speaker Diarization in their applications.
For more information and access to these resources, visit the official AssemblyAI blog.
Image source: Shutterstock