I recently posted about the Audio Overview feature in NotebookLM by Google. This time, I wanted to try something else. I created a new Notebook and imported an old digital audio recording (an mp3 file which I’d converted from an old dictaphone) of my dad and I talking about my Grandpa’s time working for the Post Office as a linesman and inspector. The conversation also included my dad speaking about his National Service in Germany. In the background my mum and husband were chatting away about other things.
The recording was just over an hour long and my aim was to see whether NotebookLM could transcribe the conversation. I prompted it to directly transcribe the audio into text and, after a few minutes processing the file, I was presented with a summary of the content of the discussion and below this was a full transcript, all ready to copy and paste into a Word document.
Screenshot showing the three panes in NotebookLM with the summary in the centre (click to enlarge)
It did very well, considering people were talking over each other at different times. At the beginning, I could tell my dad and I were looking at photographs of items which had belonged to My Grandpa Hibbitt in the First World War. When I listen back to the audio, I notice we didn’t make a great deal of sense but AI did a faithful job, nonetheless. I will, however, need to make a few corrections here and there, particularly with place names. I will also want to clean up the text to improve its readability, not the fault of AI but how we were speaking.
At the bottom of the screen you can type questions to pull out different aspects of what the program can see in your source material. There are also some suggestions to help you get started. For instance, one of the questions was ‘When was Grandpa promoted?’ It’s important to note that NotebookLM is a closed system and doesn’t pull information from the internet at large but what it does do is highlight where in the transcript the information comes from. You can either click on a number beside the relevant point to see the original transcript in the left-hand ‘Sources’ pane or hover over the number to see a pop-up of the same.
NotebookLM showing highlighted text in the Audio Transcript (click to enlarge)
You can then click ‘Save the note’ and the summary from your question will appear in the right-hand ‘Studio’ pane and will be available the next time you view the notebook.
I haven’t yet tried adding several sources at once as the purpose of this particular project was to get a quick transcript. However, the program does have much more potential than this, such as organising your family history research or your finances or anything else you wish to make more sense of. It’s your own personal notebook, after all!
