最近貪玩,把上課的部分內容製成AI-generated video供學員觀看,代替我的解說或學員的即時閱讀,下面便是一例:Fukuko 在述說當初如何開始 Team Teaching (45 seconds)。
除了表情頗真實外,她的嘴形和發音的高度配合,令人驚異。這樣下去,我快無工可做了。
Paul Sze - Honorary Professional Consultant, Dept of Curriculum & Instruction, The Chinese University of Hong Kong
最近貪玩,把上課的部分內容製成AI-generated video供學員觀看,代替我的解說或學員的即時閱讀,下面便是一例:Fukuko 在述說當初如何開始 Team Teaching (45 seconds)。
除了表情頗真實外,她的嘴形和發音的高度配合,令人驚異。這樣下去,我快無工可做了。
Padlet 的Canvas template,內有Connect to a post功能。昨天用這功能做一個五分鐘的展示,解說9個attribute與兩個概念的關係,發覺效果不錯。好處是同一個post可以connect to 多過一個概念,不是簡單的非此即彼,而且展示的過程很吸引眼球。
今早就著Jockey Club Blended Learning Project觀了很精彩的一課,還意外發現中一學生製作video (how to make egg tarts)的能力我自愧不如。除了三位同校英文老師也一同觀課外(這是很好的professional development活動),連校長也全程一小時在座,以表示對老師的支持。
我們說教育事業是生命影響生命,一般是指教師的生命影響學生的生命。但對我來說,這是相向的,教師的生命亦會被學生的生命影響。
Text-to-speech app雖然可以把输入的文字轉化成audio file,而且技術日趨成熟,但是narration 的聲音仍然要由app預設的選取。我曾經夢想有一天只要我輸入我自己聲線的樣本,TTS app便會採用我的聲音去generate audio narration。
最近一些AI app 聲稱可以這樣做,我試過Resemble AI,和Eleven Labs,結果都不滿意,完全不像我的聲線。如果你試用過這類app而效果滿意,有勞告知。
Just did another one, this time using a recording of my own voice. The fantastic thing is how the avatar's facial speech movements are fully synchronised with my speech.
https://studio.d-id.com/share?id=a9d1142e38219283fe7ea813a8dc87ec&utm_source=copy
Text-to-speech technology used to produce oral narrations which somehow sounded robotic. But like all technologies, they got better and better. Below is an AI-generated video which I created from a script. (I gave the app the script; it then gave me the narrated video.) You will see that if you close your eyes and just listen to the audio, there is no way you can tell whether it's from a robot or a human. Well, perhaps you can tell it's from a robot, because a human won't sound that perfect. (She made a slip of the tongue: peer --- observation, because of my typing mistake in the script.)
And then had a go with https://studio.d-id.com/, which is used to generate an avatar-presented video from a script that you input.
I made one from a personal biodata script.
I think such videos have to be short. Otherwise, viewers can get bored by looking at the same avatar for too long, not to mention the fact the oral narration will still sound a bit robotic.
https://studio.d-id.com/share?id=29aad4508d59bf44c8759bd0a732fa0f&utm_source=copy
Just had a go with https://slidesgpt.com/index.html.
The result is not bad, though it's unlikely it will give you 100% of what you have wanted. The picture that does with each generated slide often makes no sense.
On other other hand, each slide is accompanied with Speaker Notes!
I think it can still be of some help, as you can pick those generated slides that suit your purpose.
I went to the West Kowloon Park last night to watch the 水上煙火表演. Well, I admit I had had no idea what to call 水上煙火表演 in English, so I had t...