Comparing the podcasts produced by NotebookLM and Lettercast.ai using the same single sources (pdf or text)

Sources	NotebookLM	Lettercast.ai	NotebookLM greater by
Iconic British Designs: The London Tube Map and the London Eye	218 seconds	153 seconds	42%
	https://notebooklm.google.com/notebook/7e91a45e-e544-4447-98b3-e19889f99463	https://app.lettercast.ai/casts/0f2bd232-b44a-4983-af32-f6f4e559c66d.mp3
	647 words	375 words	72%
	178 wpm	147 wpm	21%
	GSE level: 62 - 66	GSE level: 58 - 62	7%
	CEFR level: B2	CEFR level: B2	=
Social rules in my country	314 seconds	129 seconds	143%
	https://notebooklm.google.com/notebook/d6ec3dd2-cb55-4087-b228-20e0b6de8525	https://app.lettercast.ai/casts/b5fae50f-80f1-4376-97ea-bb515483db56.mp3
	947 words	310 words	205%
	180 wpm	144 wpm	25%
	GSE level: 64 - 68	GSE level: 55 - 59	16%
	CEFR level: B2	CEFR level: B1+	+
Eat and drink – at the right time!	252 seconds	172 seconds	47%
	https://notebooklm.google.com/notebook/af091734-070b-464c-b5b5-64617d0a355c	https://app.lettercast.ai/casts/450f171b-b2fe-4d85-8c6c-e1e4fa3e57f4.mp3
	797 words	405 words
	190 wpm	141 wpm	97%
	GSE level: 62 - 66	GSE level: 62 - 66	35%
	CEFR level: B2	CEFR level: B2	=
Spanish honeymoon couple on the Titanic	267 seconds	126 seconds	112%
	https://notebooklm.google.com/notebook/39bc7cb2-54e4-407e-a6cd-24c031f7350c	https://app.lettercast.ai/casts/7f3bf733-92c1-46d3-9773-c9276b5940c1.mp3
	823 words	309 words	166%
	185 wpm	147 wpm	26%
	GSE level: 57 - 61	GSE level: 62 - 66	-8%
	CEFR level: B2	CEFR level: B2	=
Averages	263 seconds	145 seconds	81%
	804 words	350 words	130%
	183 words per minute	145 words per minute	26%
	GSE level 63	GSE level 61	3%
	CEFR level B2	CEFR level B2	=

 

There are some major differences between these two ways of generating podcasts. NotebookLM is much faster and the interface is much more sophisticated and it’s much easier to share the recordings. What’s more Lettercast only accepts PDF files as sources. NotebookLM even lets you upload audio files now!

The podcasts produced by NotebookLM are 81% longer on average and 26% faster, which means that the number of words is more than double (130% more). The levels on the GSE and CEFR scales are very similar (B2) and suitable for students at B1 and up, but not lower.

The major problem  with the NotebookLM podcasts is that they are very fast: 183 words per minute. NotebookLM allows you to play them at 80% which means the speed would be about 146 wpm, much the same as Lettercast. Sadly, Lettercast only allows you to play the podcast faster!

My suggestion is to use Turboscribe.ai to get a transcript of the podcasts which can then be listened to with the script highlighted section by section, although you have to move the page up all the time.

From NotebookLM you can download the .wav file and then upload it to Turboscribe, specify that there are two speakers and click on TRANSCRIBE. You can change the speed to 75%, which will reduce the speed to about 137 words per minute.

From Lettercast.ai It is more complicated. You need to click on View all and then on the orange RSS feed icon, find the link to the audio file which is an .mp3 file select it and select Copy and paste the link into Turboscribe, specify that there are two speakers and click on TRANSCRIBE. 

A further problem with Lettercast.ai is that now I have made 5 podcasts, I don’t seem to be able to make any more on my free account although it specifies 20 free podcasts per month.

Here is a shortlist of graded readers at about B1 for a girl of fourteen

 I was looking for contemporary stories related to women, love or music rather than classics like Pride and Prejudice.

I checked that they were all available on Come in. Everything in English! - Librería online de idiomas en Barcelona

I DO NOT get anything in exchange for adding links to these readers

Penguin Readers Level 5 B1: Boys Don't Cry - Penguin Readers

Penguin Readers Level 4 A2+: How to Give Up Plastic - Penguin Readers

 

Penguin Readers Level 4 A2+: The Kissing Booth - Penguin Readers

Penguin Readers Level 5 B1: The Pursuit of Love - Penguin Readers

Burlington Activity Readers 4^th ESO  A2+: BAR - Extraordinary Women 4º ESO - Burlington Books Spain

Burlington International Readers B1: BIR - Music Makers B1 - Burlington Books Spain

 

HOEPLI Level B1: The Beatles - Scott Wheeldon | Hoepli Editore

Oxford Bookworms Factfiles 3 B1 The USA – Oxford Graded Readers

 

How students who are over 18 can use https://notebooklm.google.com/ to create and listen to podcasts based on recordings of their speaking

 

Get your students to:
·       record themselves speaking
·       get the transcript
·       ask ChatGPT for suggestions
·       use the audio from their recordings to generate a podcast about what they said.

 

More than 10 years have passed since Vicky recorded herself talking about “The Spanish Couple on the Titanic”. It was a very good example of her speaking as you will see below. It was four and a half minutes long. Today there are many ways to record your voice and get a transcript for the recording. I used https://app.soundtype.ai  to make a transcript of her audio file. You can listen to it here and follow the transcript:

https://app.soundtype.ai/#/dashboard/9a7104fb-7cc9-4572-b957-9902dff88573

I then used ChatGPT  to ask for help with the transcript. I wanted:

•  a corrected version without any repetitions
• Suggestions on how Vicky could have said it better at her level (B1)
• And the next level (B2 or First Certificate).

It is possible to listen to it on ChatGPT as you read although there are only a few American voices to choose from.[i]

Here is the conversation:

https://chatgpt.com/share/66fed23f-e92c-800b-a659-7f6ee8f36ef5

I also used the recording itself[ii] to get an amazing podcast using https://notebooklm.google.com/  (It is very fast, about 180 words per minute.) It is also about four and half minutes long. Once again, I used https://app.soundtype.ai to make a transcript and spent half an hour adding in the little interjections like, ‘Right’, ‘Yeah’, ‘Really?’ etc that it missed. (This was fascinating because it is part of how these podcasts sound so genuine and a reminder that the person listening is rarely silent.) 

You can listen to it here and follow the transcript:

https://app.soundtype.ai/#/dashboard/d8c6b8af-b954-439a-9d4c-88d57eed5edb

The podcast is very fast and the problem with https://app.soundtype.ai is that you can only change the speed to 50%, 150% or 200%, whereas with https://turboscribe.ai/  there are speeds of 50%, 75%, 100%, 125%, 150%, 175%, 200%, 300% So, I used https://turboscribe.ai/ and said there were two speakers, and asked to get a transcript.

Here’s the link https://turboscribe.ai/transcript/share/3728980491468645852/tjT95C6NTKjQs8wKvFg5lrCITfEmshdd597wq1gvrK4/spanish-honeymoon-couple-on-the-titanic-podcast  

At first glance it seemed that https://turboscribe.ai/ had made a better job of catching all the little interjections made by listeners, but I ended up having to type in, in brackets, all the ones it had missed and there were a lot!

The question of whether the podcast is too difficult or not is not just about the speed, which https://notebooklm.google.com/   and https://turboscribe.ai/ can both solve by allowing you to reduce the speed to 80% and 75% respectively, but whether the language itself is too difficult. 

To measure that I used the two tools below:

• I used https://textinspector.com/ to rate the different texts.

Vicky’s original transcript was rated (w)    B1+    38%

The corrected version               (writing)    B2     46%

The improved version for B1 (writing)        B2     48%      [reading A2+ 29%, listening B2 48%]

The improved version for B2 (writing)        C1     63%      [reading B1+ 42%, listening C1 61%]

The podcast                          (writing)        C1     59%      [reading B1 31%, listening B2+ 52%]

 

• I also used https://www.english.com/gse/teacher-toolkit/user/textanalyzer/ to rate the different texts

Vicky’s original transcript was rated                GSE level: 45 – 49        CEFR level: B1

The corrected version                                     GSE level: 45 – 49        CEFR level: B1

The improved version for B1                           GSE level: 49 – 53        CEFR level: B1+

The improved version for B2                          GSE level: 62 – 66        CEFR level: B2

The podcast                                                    GSE level: 57 – 61        CEFR level: B2

The two tools don’t agree exactly on the levels but there is a clear progression from Vicky’s original through the other versions leading up to the podcast.

I haven’t had the chance to find out whether Vicky thinks she would have found it useful to listen to while following the highlighted transcript, but using a different podcast, my granddaughter said she understood 60%, but got a bit lost at two points in the podcast. I only played it once as I was treating it as extensive listening, which you only get one opportunity to listen to it, like in extensive reading.

I feel the podcast is sufficiently realistic with all the interjections to be a source of Comprehensible Input in Krashen’s sense of being a little above the learner’s level but not so far above to make it incomprehensible particularly because the learner already knows what they are talking about.

So, I recommend introducing this idea of students using https://notebooklm.google.com/  to generate podcasts based on transcripts of something they have said, or some of their writing. Demonstrate the idea with something the whole class has read or listened to and then demonstrate it with the recording of one student’s version of a retelling task half the class have done and the other half have listened to. Then make a video demonstrating this and give students a handout outlining what they need to do to get and listen to similar podcasts based on all their other speaking and writing.

In my next post I’ll include a comparison between https://notebooklm.google.com/    and https://lettercast.ai both of which can be used by students to make podcasts based on their writing and/or their recordings.

 

---

[i] By copying and pasting the transcript into Word in Microsoft 365 you can get a much wider range of voices to choose from

[ii] In a previous experiment I had used the transcript as audio could not be used as a source at that time.

NotebookLM from Google is the flavour of the week or month or even year

Everybody's posting examples of podcasts produced by NotebookLM but nobody really wants to listen to them because they sound samey and anyway they are far too fast for learners of English to really learn anything from. One solution to that is to slow the podcast down to 75%, but the problem remains that the content of the podcast is probably of most interest to the person who wrote the sources it is based on.

This is the way it works: you add a file or a number of related files to NotebookLM and it will produce a podcast with a male voice and a woman's voice talking about the content of this file or these files.

Learners of English

My interest is how this might be useful for learners of English (as that is the only language the podcast can be in at the moment), although you can supply it with sources in other languages. I feel that an advanced level student of English and even an intermediate one might be amused to listent to two people enthusiastically discussing what they had written or said. 

Students are likely to find it interesting and fun to hear their ideas discussed in podcast format, making it more likely they'll listen and absorb language patterns.

To test this I carried out three experiments.

• I uploaded a pdf file of a composition I had written in my C2 Catalan course and I had to imagine that I could then listen to a podcast discussing what I had written in Catalan even though, of course, it was, in fact, in English. I think I would have been delighted. The video is 7 minutes long and I found it is better to listen to it more slowly. At 75% it would take 9 minutes to view it:

• The second experiment was to use a .pdf of a composition written by a student of mine who had given me permission to use her writing, speaking and photographs for training purposes. It was a composition I had corrected and was covered in red and green ink.

Despite being covered in red and green ink, ChatGPT and NotebookLM had no problem using it. Here is the corrected version produced by ChatGPT:

You can watch the video made from the podcast recording here. I think it's best to reduce the speed to 75%

What do you think this B1 student would have made of this podcast about her composition? According to Text Inspector this is C1 57%, so some way above Dolor's Level. (Lexically, it was C2, but Readability was B1/B1+)

• The third experiment was to take a transcript of a recording made by another good B1 student and add that as a source simply by copying and pasting it into NotebookLM 

I copied and pasted various version of the transcript into The GSE Teacher Toolkit including the transcript from another tool that generates a podcast, Lettercast.ai - Turn your content into audible experiences Here is a graph showing the levels of the different transcripts:

The graph shows that the versions I asked ChatGPT to produce with corrections and improved versions for B1 and B2 should all be useful for a good B1 student to acquire some emergent language from. 

The Lettercast podcast is not much more difficult then the original and even the NotebookLM version is still only B2, a level or half a level above the student, who in any case knows a lot about the content as the podcast is based on what she said.

My next post will probably be about using the audio recording itself, which only became an option very recently.