We introduce VoiceCraft, a token infilling neural codec language model, that achieves state-of-the-art performance on both speech editing and zero-shot text-to-speech (TTS) on audiobooks, internet videos, and podcasts. VoiceCraft employs a Transformer decoder architecture and introduces a token rearrangement procedure that combines causal masking and delayed stacking to enable generation within an existing sequence. On speech editing tasks, VoiceCraft produces edited speech that is nearly indistinguishable from unedited recordings in terms of naturalness, as evaluated by humans; for zero-shot TTS, our model outperforms prior SotA models including VALLE and the popular commercial model XTTS-v2. Crucially, the models are evaluated on challenging and realistic datasets, that consist of diverse accents, speaking styles, recording conditions, and background noise and music, and our model performs consistently well compared to other models and real recordings. In particular, for speech editing evaluation, we introduce a high quality, challenging, and realistic dataset named RealEdit. We encourage readers to listen to the demos at https://jasonppy.github.io/VoiceCraft_web.

翻译：我们提出VoiceCraft，一种令牌填充的神经编解码语言模型，在音频书籍、网络视频和播客的语音编辑与零样本文本转语音（TTS）任务上均取得了最先进性能。该模型采用Transformer解码器架构，并引入令牌重排流程，通过结合因果掩码与延迟堆叠实现现有序列中的生成。在语音编辑任务中，VoiceCraft生成的编辑语音在自然度方面几乎与未经编辑的录音无法区分（经人工评估）；对于零样本TTS，我们的模型超越了包括VALLE和主流商业模型XTTS-v2在内的先前最佳模型。关键的是，模型在包含多样口音、说话风格、录音环境及背景噪音与音乐的富有挑战性的真实数据集上进行了评估，与其它模型及真实录音相比均表现出一致优越性。特别地，针对语音编辑评估，我们提出了名为RealEdit的高质量、高难度真实数据集。我们鼓励读者访问https://jasonppy.github.io/VoiceCraft_web聆听演示。