Some time ago, I found a new way of learning a language, that has become one of my most favorite onces. The method itself contains of listening to a sentences, and then trying to produce it, with all the correct sounds, correct tempo, correct lowdness, and correct pitch. I was doing that using a really well build French deck from Languages on Fire. I would just take a sentences, and try to repete it, I woudn’t be able because I wouldn’t know how to produce the sounds of the lanague, and wouldn’t even hear them clearly. Then I would go to youtube to find how the sound is produced, and how it’s different from another that sounds just like the first one. At the end I made the front side of the card as a text, and now when I review cards I have to read the sentence automatically with the same intonation as the French person in the recording doing it, and if I don’t, repeate after the native a few times the harder part, then fail the card.
Doing this, I understood that by working so much on pronunciation, on how to say it as a native person, I’ve managed to get those sentences so deep into my head that when the situation comes, I can use them without thinking at all even If I just started learning French.
That is when the idea of using Bunpro in a simmilar way came to my mind.
But that is not all. A week or two ago, I decided to try emurse.io, the website of Dogen for pitch accent training. And what I found, was that recording my own voice and listening to it, helps a lot in understanding what sounds I’m producing in a wrong way, and just by trying to say it a few times, asking LLM how to produce a cirtain sound, watching some youtube videos, I was able to correct most points I didn’t like about my pronunciation really quickly.
So now when I had those two experiances, I decided to work like that with Bunpro, and I created this little Tempermonkey script (using Gemini) that allows me to record and play my own voice directlly in Bunpro.
Script on Github
If someone is intrested in this, the points where attention is needed while doing this practice are:
- Mouth, tonge, teath position, you can find videos on youtube when you feel that you cannot produce a sound correctly. Good but limited playlist-guide on how to produce sounds. I recomend combining it with LLM, because he dones’t say anything about aspiration, and where you should feel those sounds in mouth
- Aspiration - how much air you exhale when saying one or anothe word
- Pitch - how many HZ up or down your voice goes
- How sentences connect together, how words change based on what is in front and behind
There are much more, but those are once I found usefull for me the most. Good lifehack is to ask some LLM how is your langauge fundamentally is different from the Japanese, where are they spoken (I mean deep in the moth? On the tip of the tongue? Which language is more aspirated? What are the simmilar but still different sounds? etc.)
In the future I also want to add a new functionalaity to this script, so it would be easier to listen to the Bunpro’s native recording, maybe loop the hard point, etc.
If you guys have any thoughts about all this, be free to share them!