トリコ 東京支店 Official site

Multi-variate noise suppression (MVNS) is designed to improve speech quality by effectively mitigating background environmental noise. It can be universally applied to all audio-related applications, enhancing the user experience. Potential latency: 20 milliseconds. Model size: 150KB to 6MB. MVNS Gen 1: POLQA score average improvement of 1 (0-15 dB noise).

SPEECH PRO = (ICASSP 2024) Designed to improve voice quality and clarity degraded by various real-time scenarios. Communication Specifically addresses quality degradation caused by background noise, reverberation, and other spectral coloration. Issues caused by faulty microphones and speakers, codec differences, signal volume, and interruptions due to packet loss Can be used in all audio applications to enhance signal quality Can be used to improve ASR quality in call center applications Latency: 7 ms Model size: 13 MB Packet Loss Concealment Designed to reconstruct lost packets in real-time communication of up to 100 milliseconds. Implemented in the decoder for audio waveforms, allowing it to handle any audio codec. Can be used in real-time communication to enhance sound quality Can be used to improve ASR quality in call center applications Latency: 20 ms Model size: 2.3 MB

Miamitech

Meeting: Clear communication for effective decision-making Call Center: Clear audio from nearby agents Cloud Integration: Voice enhancement to improve transcription quality. Audio AI in the cloud Smart Home Devices: Enhancing the audio experience of IoT devices Automotive Connectivity: Enhancing in-car communication and voice commands Security Systems: Utilizing voice biometrics for secure access control Telemedicine: Ensuring clear communication between patients and doctors, improving transcription Podcasting Studio: Providing studio-quality sound for content creators AR/VR: Enhancing the overall experience