Quick thread about #SearchOn22.
1⃣ I took the main video (youtube.com/watch?v=8qGV_O)
2⃣ used Whisper automatic speech recognition (ASR)
3⃣ summarized it
4⃣ asked a question about it (both using transformers)
Here is the response:
Conversation
Replying to
It's only a few lines of code, but, if you are interested, here is a Colab for you 👉 wor.ai/yt-transcribe.
Why am I interested? As the SERP becomes more visual, analyzing video content becomes strategic for many of our workflows.
Happy #SEO automation!
3
6
41
New experiment for #WordLiftNG and :
1⃣ the video this time is in German (youtube.com/watch?v=SnKd0f)
2⃣ we generate a short summary (in English)
3️⃣ we extract two questions and let the AI write the answers (these will be added to the KG via FAQ structured data)
2
5
1
1
Hi Todd, yes I will add GPT-3 as an option to this workflow - here is what we would get in this case. It sounds better while getting less into details. Thanks for sharing it.
1
Yes, interesting - not in the original text extracted by Whisper. I can guess that the summarizer tried to explain how things are becoming easier than usual with search.
1
What a great time to be alive!
This gives me so many ideas for making document specific analysis now w/ videos.
I really like the output of your question btw.
1
1
Indeed, I will keep on improving this workflow as there is so much in videos that can be exploited. Keep me posted on the analysis you'll make.
1
