Added example Podcast_and_Audio_Transcription #665
Conversation
Adds automated audio transcription using Gemini 2.0 with: ✅ Speaker identification (labeled or as Speaker A/B) ✅ Precision timestamps ([HH:MM:SS]) ✅ Music/sound effect detection (e.g., [Jingle] or [Song Name]) ✅ Clean text output with [END] marker Testing: Verified with podcasts & call recordings. Deps: jinja2, Gemini API client. Useful for podcasts, interviews, and call analysis.
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
github-actions
bot
added
the
status:awaiting review
|
Thanks @SonjeVilas, that's an interesting example. I won't have time to review it today but I'll try to do it next week. |
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
Line #1. %pip install google-genai jinja2
Since google-genai is already installed, it might be cleaner to move the jinja2 installation to the top like genai
Reply via ReviewNB
There was a problem hiding this comment.
Ok I can remove that from the code snippet.
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
Small nitpick: it’d be good to follow the string formatting conventions used in the cookbook repo—keeping lines concise and using proper indentation for multi-line strings. For example, the formatting could look like:
Template("""
Generate a transcript of the episode. Include timestamps and identify speakers.
Speakers are:
{% for speaker in speakers %}- {{ speaker }}{% if not loop.last %}\n{% endif %}{% endfor %}
...
""")
Reply via ReviewNB
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
It would be better to add an example that readers can use right away. It has to be open sourced.
Worst case, you can use the Apollo 11 audio recording we're already using in other notebooks.
EDIT: better idea, make one using NotebookLM!
Reply via ReviewNB
There was a problem hiding this comment.
Ok, I can make the notebook more readable to everyone
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
Line #47. model="gemini-2.0-flash",
Can you add a "MODEL_ID" variable like this?
MODEL_ID="gemini-2.0-flash" # @param ["gemini-2.0-flash-lite","gemini-2.0-flash","gemini-2.5-pro-exp-03-25"] {"allow-input":true, isTemplate: true}
As it would make the notebook easier to maintain in the future.
Reply via ReviewNB
There was a problem hiding this comment.
Yeah! Definitely I can make a changes
| @@ -0,0 +1,289 @@ | |||
| { | |||
There was a problem hiding this comment.
You need to fix the URL used by the button (cf. https://github.com/google-gemini/cookbook/actions/runs/14262332934/job/39989516292?pr=665)
Reply via ReviewNB
There was a problem hiding this comment.
Ok Thank you for addressing I can resolve the issue
There was a problem hiding this comment.
Hello @SonjeVilas,
That's a nice example. On top of what @andycandy already reported, and the minor stuff I pointed out, can you:
- move the notebook in the
examples/directory - add a link to it in the examples' README
- add a "what's next" section at the end of the notebook, pointing to similar notebooks (or just you preferred ones).
- run the formatting script (cf. https://github.com/google-gemini/cookbook/actions/runs/14262332934/job/39989516304?pr=665)
Thanks again!
Adds automated audio transcription using Gemini 2.0 with:
✅ Speaker identification (labeled or as Speaker A/B)
✅ Precision timestamps ([HH:MM:SS])
✅ Music/sound effect detection (e.g., [Jingle] or [Song Name])
✅ Clean text output with [END] marker
Useful for podcasts, interviews, and call analysis.