Replies: 6 comments
-
Beta Was this translation helpful? Give feedback.
-
你好,我没有勾选“包含原始文本”的选项,请原谅我使用中文来回复,我正在按你的提供的方法再次尝试进行翻译,目前还是会出现翻译好的内容,输出时格式出现错误, |
Beta Was this translation helpful? Give feedback.
-
也许可以尝试使用更小的批次,这有助于GPT保持正确的轨道。我最近注意到在翻译时字幕合并的问题更多了。将出现错误的批次进行分割并重新翻译通常可以解决问题。 圣诞节期间我应该有时间来做这个项目,所以我想尝试使用Gemini来看看它的效果如何。翻译应该是它的强项之一,但我们还得看看它遵循指令的能力有多好。 |
Beta Was this translation helpful? Give feedback.
-
好的,谢谢你的回复,稍后我会尝试的。 |
Beta Was this translation helpful? Give feedback.
-
我认为合并字幕是GPT最大的问题,我觉得它有时需要在源文中结合几行来理解意思,然后在翻译时无法将它们分开。 使用较小批次效果更好 - 我昨天翻译了一个采访,大约60行的批次在翻译中被压缩到大约20行,但当我将其分成大约20行的3个批次时,它设法在翻译中保持每一行的独立性。 如果合并行可以使翻译更好,我愿意让GPT这样做,然后调整开始和结束时间,使它们保持同步。我尝试时没能让重新同步部分工作,但我想再试一次。 |
Beta Was this translation helpful? Give feedback.
-
是的,这个比较困难,结合上下文可以使翻译更准确,确实也会让字幕句子和时间轴对不上。 |
Beta Was this translation helpful? Give feedback.
-
When translating from English to Chinese, it is required to only retain the translated Chinese content, but the original English content sentences are also generated into the translated Chinese content. In addition, the original 793 subtitle lines were reduced to 759 subtitle lines after translation, and a translated Chinese subtitle line was duplicated to generate the same two lines of content.
Beta Was this translation helpful? Give feedback.
All reactions