Wow, this is the day I’ve been waiting for - 3 months of not missing a single day transcribing Pimsleur German Levels 3, 4, & 5 into LingQ using an online “speech-to-text” app. Wow, 90 days of straight Pimsleur. The problem was that I never knew what the heck they were saying and it totally frustrated me to be listening to something that I wasn’t sure I was repeating back correctly. By using the speech-to-text app, I was able to put it all in LingQ. Finally, I feel I have gotten my money’s worth from Pimsleur. Plus the best part is I have in now in LingQ and can listen to it right in LingQ and follow along with the text. I recognize it’s an “all audio” program and that’s how they want you to use it but for me it didn’t work. I could not distinguish what they were saying. If they had broken it down slowly, the another time at normal speed (which they do only occasionally) I might have had a chance. Of course they are going to be a few errors, in the speech-to-text rendering. But usually I easily spot them and the overall benefit of “totally useless” to “now at least it got me through the program” out weighs the here-and-there transcribing errors.
Thank you Lord…
You mean you were able to play these recordings and create a transcript using voice to text software? Wow! What software did you use? How much editing was required? I wonder if it would be possible to do this with podcasts.
Hi there Mark - it does take a certain finesse to get it right. But I actually love doing it. Frankly, it’s what motivates me. If I just have to study anything I’m instantly bored. This is wrong but I can’t help myself. But if I’m entering the material, like putting it in LingQ or Quizlet or something where I get to “play” with the technology, then I love it. I’m always aware of making sure I don’t fall in the “technology trap” of doing all this busy work without actually studying/learning the material. So I’ve found a nice balance. Maybe I can make a short video for you and post it here on LingQ. It’s impossible to explain, but if you saw my screen while doing it, you’d see how its done - and as I said, I love doing it. I’m good with technology so we like what we’re good at right?
Best,
Daniel Léo Simpson
Composer
San Francisco
Interesting! Would love to see the process although if it’s that involved, may not be one for me! I am quite happy using technology to convert ebooks that I import into lessons on LingQ but it is relatively painless and generates a lot of learning material for me.
I looked on your screenshot, and as a German native I must say that the German text is not 100 percent correct …
It seems that technology can still not do it completely right.
When I OCR texts, and inevitably get errors, I find that I can catch most of them when LingQ starts giving me nonsense suggestions, and I can correct them from there. Getting “mostly there” is still very useful.
Yes, especially with old FSI manuals with umlauts typed on old typewriters for example, it’s useless - or at least all but impossible - faster to just type in text. But please note I’m not using OCR in my post above. I’m using audio-to-text. Pimsleur is playing on my computer and the app is translating it to text in front of me. I’m not pasting in anything. I then COPY when the speech transcribes and past it into LingQ. There ARE errors. For example he may say, “Sagen sie zu einem Verkäufer” "Say (such and such) to a salesperson… But it should be Sagen Sie, not Sagen sie - it didn’t catch the formal capital S. Also, I have iTalki lessons and this gives me an opportunity to have my teacher share my screen and point out the little errors here and there. But the BIG TAKEAWAY here is that without doing this, even with the errors, I would be doing NOTHING. Pimsleur would be sitting in the closet - more money wasted on unused language materials. Besides, for me, it’s actually FUN doing the audio-to-text. It’s a game for me while learning at the same time.
Daring to revive this thread. After reading the original post 3 months ago I spent some time with different speech to text services. My target language is French. Most of the time I spent was with a service by Google where I specify what language I am exposing the service to. Results were severely mediocre. I played different speeds, voices, volumes of native French but the service would only catch a few words in the beginning and then stop. Looked at services that include installing software on a computer and feeding it audio but was never worth it.
Q #1: Did you share anywhere more details of your method?
Q #2: Have you tried it with languages other than German [since then]?
The end goal for me is to have audio + text paired. It took me some time to realise that going in the other direction (text to speech) also provides the same result. In the past I have used PC client e-reader (ICE Reader) which allowed for text to [robotic] speech. It was in my mother tongue so the result was enough for my needs.
Q #3: Has anyone here used text-to-speech for language learning purposes?
My most recent dive in the text+audio world was an attempt to use my smartphone’s “Accessibility” mode to read out the text on my device’s screen. That worked well for menus but the e-book application proved to be a ‘black box’ and the assistant could not get a hold of it.
Q #4: Has anyone had any luck using this method (Accessibility on a device)?
Best regards
