Humelo Unveils Ultra-High Quality AI Voice Synthesis Tech: 24kHz to 48kHz

Voice AI startup Humelo has opened new horizons in text-to-speech (TTS) technology. They have independently developed and announced an upsampling technology that surpasses the existing 24kHz sound quality, achieving professional-grade 48kHz sound quality primarily used in movie and music streaming. Sampling rate, a key indicator in digital audio, determines sound quality. While subtle expressions are difficult with 16kHz voices from standard phones or AI chatbots, Humelo's 48kHz technology vividly reproduces even subtle breaths and vocal textures, providing a natural, human-like voice experience. This is a significant advancement that can maximize content immersion. The core technology unveiled by Humelo is 'Voice Super-Resolution Upsampling'. Remarkably, this technology can restore even low-quality 8kHz audio data to 48kHz high-resolution sound. Furthermore, its processing speed records RTFx 100, meaning it boasts astonishing efficiency, converting 100 seconds of voice data into high quality in just 1 second. Previously, the implementation of 48kHz high-quality TTS had low accessibility due to the difficulty of securing high-quality original audio data and massive computational costs. However, Humelo CEO Kwon Yong-seok stated, "We have solved the issue of delayed high-quality TTS adoption due to cost problems with our proprietary technology," emphasizing that creators and businesses can now utilize top-tier voice synthesis technology at reasonable costs. Humelo plans to lead the popularization of high-quality voice AI through this technology.