This startup offers an AI-powered transcription service accessible via API and Telegram bot, capable of identifying multiple speakers and labeling their contributions. The platform's active pattern search and self-learning capabilities improve sequential data processing for more accurate and efficient transcriptions.
Funding
Funding not disclosed
Founders
Product
Problem
Businesses require accurate and secure transcription services for multi-participant conversations, but existing API-based solutions may lack sufficient data security and speaker identification capabilities. Traditional speech-to-text services often struggle to differentiate between multiple speakers in a conversation, leading to inaccurate transcripts and inefficient workflows.
Solution
KanjuTech offers an AI-powered transcription and diarization solution designed for businesses needing high data security and precise speaker labeling. The platform utilizes brain-inspired AI to convert dialogue records into accurate transcripts, automatically detecting and labeling contributions from multiple participants. Unlike API-based solutions, KanjuTech's pre-trained AI model is deployed on-premises within secure instances, ensuring data remains private and accessible only to the user. This approach provides enhanced data security, human-level accuracy in transcription across 10 languages, and precise speaker identification even in conversations with numerous participants.
Target Audience
The primary target audience includes businesses and organizations that require secure and accurate transcription of multi-participant conversations, such as legal firms, research institutions, and media companies.
Features
- On-premise deployment of a pre-trained AI model within secure instances, ensuring high-level data security
- Human-level accuracy (3-8% WER) in transcribing 10 languages from real-life data
- Automatic speaker detection and labeling for any number of participants
- Speaker diarization with a low confusion error rate (CER) of 2.2% for conversations with 6+ speakers
- Compatibility with Amazon SageMaker for seamless integration into existing workflows
- Brain-inspired AI algorithms for improved sequential data processing and transcription accuracy