评论
打印
Researchers at Google have revealed a text-to-music AI that creates songs that can last as long as five minutes.
近日，谷歌的研究人员公布了一款能将文本转为音乐的人工智能，可以创作长达5分钟的歌曲。
The team introduced MusicLM with a number of examples that do bear a surprising resemblance to their text prompts.
该团队展示了用这一名为MusicLM的人工智能生成的一些案例，结果与他们的文本描述惊人地相符。
It has not yet been released to the public, with the authors acknowledging the risks of potential "misappropriation of creative content"
MusicLM目前还没有向公众发布，作者承认潜在的“创造性内容会被盗用”的风险，
should a generated song not differ sufficiently from the source material the model learned from.
即生成的歌曲与模型学习的源材料没有足够的差异。
The researchers claimed their model "outperforms previous systems both in audio quality and adherence to the text description".
研究人员声称，他们的模型“在音频质量和对文本描述的依从性方面都优于以前的系统”。
They said it is a "hierarchical sequence-to-sequence model for music generation"
他们表示，MusicLM是一个“分层的序列到序列音乐生成模型”
which uses machine learning to generate sequences for different levels of the song, such as the structure, the melody, and the individual sounds.
它通过利用机器学习，为歌曲的不同层次生成序列，比如结构、旋律和单个声音。
The model is trained on a large data set of unlabeled music, along with a music caption dataset of more than 5,500 examples, which were prepared by musicians.
该模型是通过一个大型的无标签音乐数据集进行的训练，同时还有一个由音乐家准备的超过5500个示例音乐文本数据集。
This dataset has been publicly released to support future research.
该数据集现在已公开发布，以支持未来的研究。
