Abstract: Provided are a synchronization method and apparatus for audio and text, a device, and a medium. The method includes: determining a plurality of first text segments for audio conversion and a second text for reading display, in which the plurality of first text segments and the second text are from an initial text; converting the plurality of first text segments into audio segments, to obtain a first mapping relationship between the first text segments and the audio segments; performing matching on the first text segments and the second text, to obtain a second mapping relationship between the first text segments and second text segments in the second text; determining the second text segment synchronized with each of the audio segments based on the first mapping relationship and the second mapping relationship.

Abstract: A weakly paired image style transfer method based on a pose self-supervised generative adversarial network, relating to the field of image processing. The method is suitable for style transfer of weakly paired images, different styles of pictures having certain overlap are used to perform model training of an adversarial neural network, so that the model is insensitive to poses and focuses on style learning, and therefore, in an actual application process, a source style can be converted into a target style, but a pose is kept unchanged. In addition, in the model training process of the adversarial neural network, a differentiable pose solver capable of estimating a relative pose of any two images is introduced, a phase correlation algorithm is optimized to be differentiable, and the phase correlation algorithm is embedded into an end-to-end learning network framework to achieve pose estimation.

Abstract: An audio conversion method, an audio playing method and an apparatus, the method including: receiving an audio acquisition request corresponding to a target chapter (101); in response to an absence of an audio file corresponding to the target chapter, segmenting the target chapter to obtain a plurality of text segments (102); generating an audio file corresponding to each of the text segments, and determining identification information of the audio file based on a typesetting order of each of the text segments in the target chapter; storing the audio file corresponding to each of the text segments, and generating an audio list based on file information of the audio file corresponding to each of the text segments and the identification information of the audio file (103); and determining an estimated total audio playing duration corresponding to the target chapter, and sending the audio list and the estimated total audio playing duration to a user terminal (104).

Abstract: A method for fabricating a semiconductor structure having an enhanced hole linear Rashba spin-orbit coupling effect includes: providing a substrate; and growing a germanium quantum well on the substrate. A silicon atomic layer is inserted at an interface between a well and a barrier of the germanium quantum well. The silicon atomic layer includes one or more monolayers.

Abstract: A fabrication method for a semiconductor structure with a hole spin qubit includes: providing a substrate; growing a germanium quantum well on the substrate, in which the germanium quantum well is an inclined quantum well structure grown in a [110] direction, and the germanium quantum well is grown by a complementary metal oxide semiconductor process; and fabricating a two-dimensional gate-defined quantum dot in the germanium quantum well.