Geely Auto has recently achieved a significant breakthrough in the field of speech synthesis, with its self-developed HAM-TTS large model outperforming the industry benchmark VALL-E, attracting widespread attention in the industry. This AI large model, named "Xingrui," has achieved significant improvements in key indicators such as pronunciation accuracy, naturalness, and speaker similarity. The HAM-TTS model utilizes token-based zero-shot text-to-speech hierarchical acoustic modeling technology, greatly enhancing the user interaction experience in intelligent cockpits. Under the same conditions of 400 million parameters, HAM-TTS...

微软近期发布了一款零样本文本到语音（TTS）模型VALLE-2，该成果在技术界引发热议，被认为是TTS领域的里程碑。VALLE-2的创新之处在于其零样本学习能力，仅需一小段陌生语音样本，便能模仿声音说任意文本，实现了惊人的即时模仿和编码器分组。它还改进了重复感知采样，提高解码稳定性，并简化了数据需求。在主观评分和客观指标的测试中，VALLE-2超越了前代模型VALLE，几乎达到人类语音水平。然而，强大的语音模仿能力也引发了对Deepfake技术滥用的担忧，特别是关于伦理考量和潜在风险，如模型对非英美口音处理能力及计算效率的改进空间。当前，微软仅将其作为研究项目，并强调了合成语音检测和授权机制的重要性。未来在语音助手、内容创作等领域的应用值得期待，同时需要关注技术的伦理使用与风险防控。更多关于VALLE-2的信息可访问项目页面：Microsoft官网。