Wish my Canto lesson audio were clearer

[querido]

This post is not a question but is just what's on my mind right now.

 

I have some well-regarded Cantonese learning materials that should take me to intermediate. Can't stand the audio! :-( 

 

I tried re-recording it with my teacher but it's technically problematic (echos from hard surfaces, needs EQing, etc), and she doesn't speak perfectly either. 1/3, 1/4 of sentences need to be redone, and it's too complicated to make fine judgments about what is perfect enough and what is natural deviation (and thus should be left in there).

 

**

What I want to hear, and how I would like to sound myself, is the vigorous staccato I hear from newscasters and also from lots of regular people all over youtube, etc. You know this sound, right?

 

Now I'm annoyed and discouraged with these materials. I had imagined just letting it play and flying with it! :-(

 

**

I'm also working through CantoneseClass101 with one of my native speaking friends. We've done over seventy lessons and will reach their "intermediate" in about six weeks and I've flashcarded everything. The audio there is good, just a little slow (which is not a problem because this would be easy to adjust with Audacity, see?). But guess what? As I work through these lessons my friend #2 makes major changes to the text. This is ok because I double-check it with friend #1 who usually agrees with the changes, but it means that the audio from CC101 no longer matches the text! :-(

 

**

Someday, one fine day, hopefully not in the sweet by-and-by, I hope to switch to native materials with subtitles and audio that's delivered as I wish to hear it. Don't care if the subtitles are SWC. But I really need to cram all of the above, because of its practical vocabulary, in the next six or seven months.

 

Ok, thanks for listening.

 

Edit:

The proverb says "if you want it done right you must do it yourself". I have friends who could critique, and it would be productive effort. I did well at this in Mandarin, but I was never fl**nt in Mandarin, and doing this would draw me toward printed words again, when I was hoping for something I could just play. I'm annoyed because I've already thought this through: no amount of reading aloud will get me where I want to go in Cantonese. I could record it, and THEN play it ad nauseum. But I wonder why I should be doomed to listen to myself. :-( Can't be ideal. :-(

Ok: I could record everything in both of these sources (vocab would be about 3-4K words), probably learning it pretty well as I go, and then not listen to THAT, on repeat, but immediately switch to native materials. 

[Shelley]

Post removed to avoid confusion.

[querido]

Post was removed to avoid confusion.

[Shelley]

Again, post removed to avoid confusion.

[querido]

Again, post was removed to avoid confusion.

[wushucrab20]

For what its worth I think that whether you use standard materials or not, it will all be fine for helping to bring you to the intermediate level. Just listen to a lot of things, preferably stuff you are interested in and even if the speech is not so perfect it will help you understand perfect speech when you hear it anyway, and those adjustments aren't very difficult to make later.

[querido]

Thank you wushucrab20. I admit that it's good practice for understanding real-world speech, and I know I'm making progress because items get easier over time, etc. I also admit that I'm crying under this workload. 

 

With Anki's AwesomeTTS plugin I have been adding "second opinions" when necessary from Google TTS (just takes a few seconds). I hesitate to say how often, but a big fraction of the time it is better than the provided audio. It makes mistakes too, many of which are easily correctable (I've collected a list of substitutions). So, I have the provided audio, re-recordings by my teacher, GoogleTTS, and myself. Didn't forsee and didn't want the complication.

 

Now stubbornness - my greatest virtue - could refuse to accept the provided audio, or could insist that it's good for me. We'll see.

[carlo]

I've been thinking recently about how to go about finding Cantonese audio materials that would help me improve. My favourite activity at the moment is watching online videos and transcribing them -- very slow going, but excellent for listening comprehension.

 

I was talking to a friend here in HK, who's worked as a newscaster, about the possibility of having him record some audios for me. What kind of audios would you find useful in your own study, Querido? I personally have an aversion to most "recordings for second-language learners" -- deliberately slow, artificial enunciation that doesn't match the kind of language I normally hear all around me.

 

The best audio materials for me would probably be native-speed, with interesting content, and very accurate transcriptions so that I can use the transcriptions to revise and retain what I am hearing. I'd love to have your thoughts on this -- maybe I can get him to record something that can be of use to others too.

[querido]

You're asking a general question about language learning materials. We have a lot of threads about that. My contribution is just a common-sense warning not to let your worst skill fall too far behind. In my case (in the past, in Mandarin) it was listening comprehension. But I could understand my podcast audio and could read aloud and understand my own voice just fine; it was one aspect of listening comprehension in particular which, since you asked, I'll talk about again.

 

I need someone or something to feed me words I already know but continuously rearranged to simulate original conversation, then wait a moment for me to construct replies (when I will of course use words I already know), and then use my replies to continue along a conversational thread. My experience so far is that people seem to be unable to do this. I believe I could do it if I were teaching English, but sorry, my teachers either can't do it or don't understand the problem. (I have a theory about why this is mysteriously difficult for them.) So, as I said, I should learn Cantonese and then talk to myself; I have a feeling that that is exactly what will happen someday! :-)

 

I think we'll have implantable chips or something before the producers of language learning materials understand the problem and can connect the tools (which probably already exist) together to do that. Long, long ago we had amusingly good artificial conversational partners, in text only. Now we have computers winning at Jeopardy! against humans. Somewhere in between (and thus affordable and do-able now) might look something like this: audio > speech to text (note 1)> basic text "artificial intelligence" (code for which you could probably find on the Internet) told to stay within a given vocabulary (note 2) > text to speech > audio.

 

Note 1: If the speech to text demands perfect* pronuncation, and is limited to the given vocabulary, no problem in either case.

Note 2: Like this. This much has been possible for decades:

- Good {time of day}.

-- Good (same time of day). How's the weather?

- The weather is {adjective that can describe weather}. Did you eat {name of a meal} yet?

-- Yes, I had {name of a dish}. Hey, is that your {noun}?

- Yes. It's a {color that that noun can be} (same noun).

-- I once had {number} {color that that noun can be} one(s). (If I say, knowingly, in Cantonese, "I once had a thousand purple cars", who cares?)

- And oh, by the way, won't you tell me about your mother? :-)

-- Ok, if you'll open the podbay doors then I promise to tell you about my mother. :-)

etc.

 

**

Here's a quickie solution (but it still doesn't touch the part where I need to generate the content and then enjoy their reply to what I said):

The people who are already in the business of producing podcasts could, every x lessons, produce a sub-series of podcasts of original conversation using no new words. No need to explain new words or new usages. And I can imagine prompts for replies (like e.g. Michel Thomas, right?). Considering the simplicity of a beginners capabilities - which must be prompted and then replied to in a simple way - I don't know why this must be rocket science; when I speak English to my teachers I simplify my sentences and my vocabulary. It has become automatic. I check their face for understanding and rephrase as necessary. You need someone to whom conversation is an engagement with someone else's mind (not an exchange, back and forth, of something you heard somewhere, whether or not it is connected to what the person said). It's an ugly truth, and a blot on humanity's reputation, and why I don't want L1-speaking friends, that such people are so rare. Actually, most people can do it when they're suitably motivated, but most of the time they just don't care enough (about the other person). 

 

Here's a better one: Find, somehow, a native speaker who can converse while staying within a given vocabulary. Conduct hour-long sessions in L2 only. 

 

The "forest-for-the-trees" of language learning is "lovely, dark and deep". "But in the long run, there's still time to change the road you're on". :-)

[skylee]

Re #8, how about listening to RTHK? It is an HK Government radio station. There are no ads, and RTHK is known for not toeing the official lines. I listen to RTHK2 every morning from 0550 until I leave home for work at about 0800. The hosts all speak in perfect natural Cantonese of native speed. The shows are informative and interesting IMHO.

I listen to it live via the app "tunein radio" when I am abroad. You can also listen to the recording of the shows on its official website.

[querido]

Look at their podcasts too! http://podcast.rthk.hk/podcast/

[querido]

Anyone interested in hearing it?

Original: Sample 1.wav

Original EQd at straight 3db/oct from bottom to top (reducing bass) and renormalized: 3db per oct.wav

 

Coincidentally, that correction is very close to an inverse RIAA eq curve. Is that interesting to anybody?

Here is the original with an RIAA eq curve subtracted out: Inverse RIAA.wav.

 

Anybody strongly prefer the corrected ones? I do! :-)

[Shelley]

RIAA equalization is a specification for the recording and playback of phonograph records.

 

And yes that is sort of interesting to me, as i am familiar with this sort of thing, having worked in the audio business for the last 38 years.

[querido]

Nice to see you Shelley.

So it would be a mistake to use it, right?

Does anyone else think the original is annoyingly thick in the bass? I admit it could just be hearing loss.

[carlo]

Thanks for the helpful reply! I understand much better now what you are looking for. And yes, I listen/ watch RTHK every day (currently going through a very nice show called 窮富翁大作戰, also available on Youtube, but I'm going off on a tangent so I'll stop).

[Shelley]

Not sure it would be a mistake but maybe not usual, it all depends on what it sounds like to you.

 

Audio is a very subjective thing, what sounds good to you may be unpleasant to me. When I am asked to recommend HiFi speakers I always say listen to as many speakers in your price range as you can and buy the ones you like the best regardless of make or price, the best for you may not necessarily be the most expensive.

 

Remember also that lots of things influence the sound you get, not just the speakers, the size and shape of the room, high ceilings, carpets or not, soft furnishings and so on.

 

I prefer the tone of the Sample but the other two are clearer for learning purposes, I wouldn't want to listen to a story with those two eqs but good for clarity.

[querido]

Thanks.

 

I think EQing the original files will be good enough. I'm sorry about all the agonizing above.

 

Edit: If you do Audacity/Analyze/Plot Spectrum on these files vs. e.g. GoogleTTS, you can see it too.