HCS-5348/80 Intelligent Minutes of Meeting Software, For HCS-5300 Series Conference System

Based on Taipower's TAIDEN professional conference speaking system, the industry's advanced intelligent speech recognition technology is introduced to create a system that can speak (meaning that the speaker has professional speaking equipment) and listen (meaning that professional speech-to-text facilities are equipped for meetings, in real time) An intelligent conference speech and discussion system that listens to "speech content" and is visible (realizing meeting content information and data visualization). In addition to realizing fast, efficient, accurate and secure meeting information records (such as meeting minutes, training records, real-time speech subtitles, interviews (recording transcription, real-time recording of court hearings, etc.), but also to open up the barrier between unstructured data such as conference audio and video materials and text-type structured data, and convert unstructured massive conference audio and video materials into structured conferences. Data, while effectively improving the efficiency of storage, management, and use, it also provides a basis for later processing of meeting information and data. The latest Taipower conference system management software embeds the speech-to-text technology module developed by Alibaba Damo Academy Speech Laboratory, and cooperates with Taipower's independently developed digital audio transmission protocol, which contains patented technology (Patent No.: 200620054419.6), supporting up to In a meeting where up to 8 people speak at the same time, the voices of different representatives are separated to realize text output with the speaker's name and speaking time stamp. The recognition rate is as high as 98%, and the real-time voice delay is ≤100ms, which greatly improves the efficiency of conference management. . The output meeting minutes can be retrieved by name or time, making it easier to organize and archive them after the meeting.

Features

• ASR automatic speech recognition technology developed in conjunction with Alibaba Cloud, with a speech transcription accuracy of ≥98%
• Supports real-time audio stream transcription and historical recording file transcription
• Speech transcription engine deployed offline, data is more secure
• Supports deployment in conference center cluster or local conference room
• Conference center cluster deployment can meet the needs of parallel speech transcription in multiple conference rooms
• Fully compatible with TAIDEN all-digital conference system and TAIDEN digital infrared wireless conference system
• Mature voice hardware solution
•  CD-level sound quality, improved accuracy
•  Taipower TAIDEN conference system’s multi-microphone independent output technology can achieve hard separation of character microphone voices
•  Multiple microphones can speak and transcribe at the same time
•  Ensure the accuracy and integrity of speeches
• Industry-leading ASR engine
•  Industry-leading acoustic model algorithm
•  High recognition rate, high accuracy, high performance (3x computing speed, 3x response speed, 3x cognitive model)
• Massive voice data accumulation
• Full-granularity corpus self-learning, minute-level model training speed
•  Hot word self-learning: learn the name, basic and other proprietary information of participants through word self-learning; hot word optimization takes effect within 10s
•  Text self-learning: through word self-learning Learn from texts such as speech drafts or conference manuscripts; it has sentence-level and chapter-level text optimization capabilities, and the hot word
• optimization takes effect within 2 minutes
• .  Behavioral self-learning: According to the user’s usage habits, the modified information is recorded and learned, and the application background will update the user The modified text is added to the word self-learning. If the text is mentioned again during the meeting, it will be automatically recognized correctly.
• Powerful post-text processing
•  Based on a strong deep learning foundation and a large amount of data accumulation, the recognized text is processed to enhance the readability of the text and facilitate subsequent text understanding and analysis.
• Rich conference application functions
•  Real-time voice transcription: real-time voice transcription, hot word optimization, role separation, support for multi-language and dialect transcription, etc.
•  Historical voice transcription: fast and efficient, 1 hour of audio transcription can be completed in 5 minutes
•  Meeting management: speech shorthand, text editing, meeting audio and video, positioning and transcribing content review and listening, meeting information fusion, instant generation of meeting minutes, recording file transcription, etc.  Upper screen
• display: real-time on-screen, video superimposed subtitles, Multi-terminal display, etc.
• Host cross-retrieval of audio and transcribed text
• Check the transcribed documents and minutes after the meeting
• Suitable for meetings, speeches, interviews, teaching, court hearings, etc.