I got an email from Transkriptor inviting me to try it again saying that it was now "free."
So, for the first time in just over a year, I tried Transkriptor again. This is what I found with my 'trial account', where I had plenty of minutes' credit left:
Easy to edit out errors in the transcript
Easy to export the transcript without speakers and timestamps on the app
Easy to paste transcript into ChatGPT and then add 9-step prompt at the beginning
After that, it was all down to the ChatGPT app and everything worked just as well as with Turboscribe.ai
So I made a video of the recording playing on Transkriptor and then of the 9 steps on the ChatGPT app on my Android phone and posted it on my YouTube channel. I have since deleted the video as I subsequently discovered this:
If you want a 'free account' you need to give your credit card details, which I wouldn't recommend and haven't done, so I will never discover whether you get 30 minutes every day split among any number of recordings or just one transcription of up to 30 minutes. I suspect the latter.
If you have a 'trial account' with Transkriptor, there are limitations on the length (5 minutes) and percentage (80%) of your recording that will be transcribed, so I can't really recommend that. The workaround is to keep recording anything whatsoever for a quarter of the time after the time you have recorded!
I did the webinar more than a month ago and on and off in the last month I've been trying to edit the video that I made using Zoomto prepare for it. The video had to be in two parts because my free Zoom account limited me to 45 minutes.
There were a number of problems that I had to solve:
I must have coughed at least twenty times
Some of the audio I played was too quiet
I needed to add some text to suggest viewers pause the video to do the different tasks, or to explain things I hadn't remembered to explain.
I wanted to add subtitles in English, Vietnamese, Kazakh and other languages
I'm still not sure whether to post just one long video or to break the video up into four or five linked videos. Maybe I''ll do both but for the moment it's just one with chapters
Coughs
I have always used Microsofts Clipchamp to edit videos and I discovered that there are two ways to remove the coughs: you mark with the scissors just before and after the cough. Then you can either turn off the sound in that very short segment of the video or delete it. Deleting it was my first choice as it shortened a very long video.
Too quiet audio
My first approach was to mark the segment where the audio was playing, turn off the sound and add an additional audio track in its place. I learnt through trial and error that another way to solve the problem is to mark the segment and then cick on 'Audio' on the right ansd pt the volume up much higher for that segment.
Adding a text box
I found that there was an area on the right of the screen under the next slide and my face, which was always empty and where it would be best to add a text box. I had to make my instructions or comments very brief.
Subtitles in Vietnamese and Kazakh
I exported the .srt file with the subtitles in English from Clipchamp and uploaded them to YouTube. I then asked YouTube to provide subtitles in these two languages as well as in the languages in which I could check that the translations were correct: Catalan, Spanish, French and Italian.
Here, finally is the 75-minute long video with the following chapters:
00:00 Intro + using Turboscribe.ai and ChatGPT to get feedback on speaking
08:45 Demonstration of the nine-step prompt to get feedback on speaking
28:54 Now you do it - a hands-on task as if a student wanting feedback on their speaking 58:41 Suggestions for communicative speaking tasks for students to do in and out of class
1:11:47 Outro + websites that help your pronunciation, Facebook groups on edtech
I learnt about this possibility from Joe Dale on LinkedIn and Facebook and immediately wanted to use it with this recording made 10 years ago by Vicki, one of my Intermediate students.
She was an excellent B1 student, as I think you will agree when you listen to her in the video, so I suggested to Gemini that it could make a StoryBook based on the transcript of her recording and I suggested it could be at B2 as that might help her be exposed to some new comprehensible input. I discovered later that Gemini was very happy to explain any new vocabulary the student asked about.
I have used five different tools in the past to create podcasts based on this same recording, some of which I have made videos of. I also made a video of me being interviewed by ChatGPT voice about the story following a suggestion from Daniel Martin. I'll add links to these at the end.
The idea behind all of these ideas is for students to use their own content, spoken or written, to generate new content at a slightly higher level. This works very well with B1 students, who can probably handle texts on familiar content at B2, but it may not be so useful for lower levels as the sophistication of the language and the speed of the audio will probably be beyond them.
Here is Gemini's StoryBook based on what Vicki said when retelling the story of the Spanish Couple on the Titanic:
Here are some links to two podcasts and one interview about the story retold in the same recording:
I usually post about English Language Teaching, but this is different.
The other day I asked Copilot, "What address could this be from 1911 It's Southend Road, maybe number 5?" (accompanied by a hand-written address)
Copilot couldn't help with transcribing the address.
I then asked, "What was Beckenham like in 1911?" and got a lot of information.
I asked Copilot, "5 Southend road was very different then, I imagine" and got more information including, "Back then, Southend Road was lined with large Italianate villas built in the mid-1800s, part of Beckenham’s transformation into a leafy commuter haven." I asked a few more questions and then I said, "Ideally, I'd like a picture of 5 Southend Road in 1911." and Copilot said, "That would be a gem to uncover, Chris!" and suggested some places where I might try to find one.
Then Copilot asked, "If you’d like, I can help you visualize what the house might’ve looked like based on architectural styles and street descriptions from that time. Want me to sketch a period-style interpretation?"
The 'sketch' is amazing and totally unexpected. I've downloaded it and added a caption that says, "Copilot sketching how it imagined 5 Southend Road (was) in 1911" to make sure no one assumes it is a picture of that house.
The moment I heard this I wanted to try it out for myself so I tried it with the video on YouTube I had made when rehearsing for the JALT/EUROCALL joint webinar.
Here is the video of the rehearsal:
This is the summary created by NotebookLM:
Summary
This webinar rehearsal, led by Chris Fry, introduces educators to a method for providing instant speaking feedback to language students using AI tools. The primary focus is on Turboscribe AI for transcription and ChatGPT for comprehensive error analysis and improvement. Students, aged 13 and over with parental consent, can use free accounts to record their speech, receive a transcript, and then engage with a nine-step AI-powered feedback process. This process incrementally guides them from identifying and correcting basic errors to generating and listening to increasingly sophisticated versions of their spoken output, facilitating a gradual improvement in their language proficiency.
Here is the video overview created by NotebookLM based on just this one source:
This source introduces an innovative, free two-part AI system that acts as a personal language tutor, designed to help learners achieve fluency in any language. The system addresses the common problem of lacking immediate feedback when practicing speaking independently, enabling users to receive instant, detailed feedback on their spoken language. This "AI language tutor" combines a transcription tool, like Turboscribe AI, to accurately convert speech into text, with a language model, such as ChatGPT, to provide comprehensive corrections and improvements. The process unfolds across three progressive levels: firstly, fixing obvious errors in grammar and structure; secondly, polishing phrasing to ensure natural, fluent expression; and thirdly, upgrading vocabulary to enhance expressiveness and sophistication. By systematically addressing these areas, the AI not only corrects mistakes but also significantly boosts a learner's confidence and control over their language acquisition journey, working for over 90 languages.
I then wondered how the video overview might change if I added two more sources.
This source outlines a nine-step process for refining transcriptions, likely from an AI tool like Turboscribe, by systematically improving their quality. The core purpose is to elevate a text from its initial flawed state to a highly sophisticated version, moving through incremental levels of linguistic complexity and coherence as defined by the GSE and CEFR frameworks. Each step involves identifying and correcting errors, then progressively rewriting the text to incorporate more advanced grammar, richer vocabulary, enhanced logical structure, and greater detail, with all changes meticulously marked to show the evolution of the transcription.
A pdf showing the different steps and how to introduce them in different stages:
Summary of source by NotebookLM
This document outlines a nine-step process designed for improving transcription quality by systematically refining text. The methodology progresses through several "stages," beginning with identifying and correcting errors in an initial transcription. Subsequent stages focus on enhancing the language and structure of the text, first by rewriting half a CEFR level up with improved cohesion and vocabulary, then by further elevating the complexity, detail, and logical flow of the prose. The overarching purpose is to provide a structured approach for achieving increasingly sophisticated and error-free transcriptions.
Perhaps surprisingly the video overview based on three sources was actually shorter:
This source outlines an innovative method for transforming a personal device into an AI language tutor, offering immediate and tailored feedback on spoken language. It details a system leveraging two free tools, Turboscribe for accurate transcription and ChatGPT as the expert AI coach, to help learners progress from basic corrections to sophisticated expression. The core of this method is a "magic prompt" – a series of instructions that guides the AI through three stages: fixing errors, improving natural style, and levelling up language to more advanced forms. This personalised approach aims to build confidence and enhance learning by providing instant, customised feedback based directly on the user's spoken input.
Last Thursday I had the pleasure of addressing about 250 participants in a workshop about how students can get feedback on their speaking whenever they use Turboscribe to record themslves.
I started off using a DirectPoll survey to find out whether participants were equiped to participate in the workshop. Only 120 of the 250 people present responded.
So I suggested the people who hadn't got an account with Turboscribe.ai should get one as soon as possible.
Fortunately, nearly everyone had the ChatGPT app on their phone. I explained that the app was much better than the browser version when it comes to listening to texts read aloud unless they were using the Edge browser, which also allows you to select text and ask for it to be read aloud.
I explained that anyone using their phone to follow the webinar might have problems using the same device to also record themselves and use ChatGPT.
The question I didn't ask, and regret not asking was whether participants had a QR code reader or not, but of course that would only have worked if someone had put the short URL in the chat.
Nearly two hours later I asked participants to evaluate the ideas I had been putting forward and they had been using as if they were students. This time there were only 32 responses, but they were overwhelmingly positive:
Finally, I invited everyone who was still following the workshop to join my WhatsApp group called 'Speech to Text in Language Learning and Teaching'. Only 18 people joined, which was slightly disappointing.
An improved 9-step prompt
The main difference between this version of the webinar and previous versions was that I had improved the 9-step prompt to try to guarantee that there would always be a steady progression in level between the original transcription and the corrected version and the three improved versions. Here is the revised prompt:
Remove timestamps.
Decide the level on GSE and CEFR (+ for half bands) but don't reveal it.
After each of the steps 1 to 9, wait for me to say "next."
Write in continuous prose.
Answer questions at the learner's level.
Steps:
1 Identify errors in bold and enclose them in brackets
2 Correct errors, showing corrections in bold next to (errors)
3 Write the corrected version without mistakes
4 Rewrite half a CEFR level up from the version in step 3 but more logically structured, with better cohesion, more complex grammar, richer and more idiomatic vocab, and more details, marking changes as in step 2
5 Write the changed version
6 Rewrite one CEFR level up from the version in step 5, improving as in step 4, marking changes as in step 2.
7 Write the changed version
8 Rewrite one level up from the version in step 7, more complex, even better linked, even more detailed, even more logically structured, marking changes as in step 2
9 Write the changed version
Here is a QR code and short URL that link to the prompt so that you can use any slight improvments I make in the future.
I suggested that teachers and even learners could adapt the prompt to suit their needs. For example, leaners aiming to improve their IELTS speaking scores could ask ChatGPT to use this scale rather than the GSE and CEFR scales. I reminded everyone that there is a limit of 953 characters including spaces when making changes to the Custom Prompt to be pasted into Turboscribe.ai
Setting up Communicative Speaking Activities
A second difference between this version and previous versions of the webinar is that this time I had added a section with suggestions about how to set up communicative speaking activities for students in class and for homework. These were taken from my presentation for a Google Hangout with the late Vance Stevens back in 2016.
Recordings of the webinar
I recorded a rehearsal for the webinar, but haven't had time to edit it yet, but when I have, I'll share it on my YouTube channel Chris_Fry_Barcelona - YouTube
TESOL Simple Education have already shared the edited recording of the live webinar. The link is here. You will need to ask for a translation of the page into English, create an account and click on the Admissions button.
How good is ChatGPT at writing at specified CEFR levels?
I had always used Pearson's Text Analyzer to get a grade for the different versions of transcripts produced by ChatGPT, but was slightly suspicious that grades rarely went higher than B2, so I compared these grades with the ones assigned for 'writing' on TextInspector. They were often higher, particularly at B2 and above, so I used an average of the two grades as in the table below.
Using the average (and the sums) convinced me that the progression was useful even if ChatGPT isn't perfect at writing at the levels demanded of it. ChatGPT is 'good enough'!
I have created a new version of my shortest video outlining the nine steps I have recently recommended for learners to get feedback on their speaking. This version is optimized for mobile phones and includes subtitles made using ClipChamp, which were very good and only needed me to add in the punctuation.
