Mr_Future

> indeed bro，i tried recording in CSGO and do a comparison,its obviously downgrade the audio quality,so i wouldnt really use this version until the problem fixed

> This is happening due to the API's behaviour of capturing audio render streams from the child processes of the process that is specified in addition to the audio of...

> Same bad audio for me that sounds muffled but it only does it occasionally ru using win11?it dont happen usually?

> 这很正常，LARGE 模型会比 MEDIUM 模型耗时多得多，我也这样 V3最新的模型实测是只能加载，并不能正常用，还是得等作者更新

> What @Jiang10086 says translates more or less to "it is normal that large model is slow" if i get that correctly? Well, in this case we face the V3...

> @MrFutureV the N46 Jupyter Notebook you linked is cool but after all it just uses the original openai whisper python project so this is not directly useful for us...

> Still for anyone interested to test it out, here my custom version that **ONLY WORKS with the V3 model,** not with any other. This is WhisperDesktop for GUI users...

> @emcodem 的方法很有用， 使用whisperConstMe_V3_test.zip能解决这个问题（runFullImpl: failed to generate timestamp token - skipping one second） 然而还是没办法避免V3这个模型目前的输出重复问题，还有转写准确率也不如之前的那么高，还是看openai以后的更新吧

rt，出现过两次这种情况