High-speed speech synthesis can be used as an intelligence amplification hack

[lirazsiri]

I usually only lurk on the MobileRead forums but I signed up to share the interesting experience I've been having reading hundreds of books at gradually increasing text to speech rate. It turns out if you put yourself through a kind of training regimen you bump into the upper limit of speed supported by modern text to speech engines before you bump into the upper limit that your brain can be made to handle. Plus, there seems to be an unexpected intelligence amplification effect.

http://www.turnkeylinux.org/blog/int...text-to-speech

It's a bit counterintuitive but I'm hoping there might be a few eBook fans on this forum willing to give it a try and see for themselves.

[HarryT]

"Intelligence amplification effect"? Can you provide "before and after" IQ test results to demonstrate this?

[fjtorres]

The speedup comprehension effect is old news.
With pitch control it is possible to speed up audio recordings (or TTS) by a lot without impacting comprehension. Humans can process speech much faster than anybody can speak.

From there to intelligence boosts *because* of the faster data flow is a tough point to make, though. Hard to even construct a proper test to modern standards.

The correlations that have been found in experiments going back to the 50's actually point to an opposite causality: higher intelligence leads to a higher speedup limit to comprehension.

http://books.google.com.pr/books?id=...0pitch&f=false

[rkomar]

Unfortunately, the PocketBook devices have a brief pause after each sentence for the display to highlight the next line, and I think that effectively limits the speed. However, I bet a similar thing can be done with audiobooks, if you have a playing app that can change the tempo at which the file is played. I tried with a podcast, and found I could listen to it comfortably at a 50% higher tempo. Twice the tempo was too hard to follow, though, right off the bat.

[fjtorres]

Originally Posted by rkomar

Unfortunately, the PocketBook devices have a brief pause after each sentence for the display to highlight the next line, and I think that effectively limits the speed. However, I bet a similar thing can be done with audiobooks, if you have a playing app that can change the tempo at which the file is played. I tried with a podcast, and found I could listen to it comfortably at a 50% higher tempo. Twice the tempo was too hard to follow, though, right off the bat.

Yes, you need to tweak pitch lower to go really high.
There used to be old-style tape recorders with pitch adjustment for use to record material for transcription.

[rkomar]

Originally Posted by fjtorres

Yes, you need to tweak pitch lower to go really high.
There used to be old-style tape recorders with pitch adjustment for use to record material for transcription.

The tempo adjustment in the sox play program keeps the pitch the same. Are you saying that it has to be pitched even lower than the original to be comprehensible at high tempos? Or were you saying that the pitch needs to be adjusted down if the speed of the playback is increased (rather than the tempo)?

[fjtorres]

Originally Posted by rkomar

The tempo adjustment in the sox play program keeps the pitch the same. Are you saying that it has to be pitched even lower than the original to be comprehensible at high tempos? Or were you saying that the pitch needs to be adjusted down if the speed of the playback is increased (rather than the tempo)?

It needs to counter the "squirrel voice" effect at a minimum.
Going lower helped in many cases but not all. Listener dependent, which is why its not in general use.

(This is really old stuff so I don't recall all my sources or details. One of them was a video that showed a female voice getting pitched to male tones.)

[latepaul]

I think a lot of what the OP describes as "intelligence amplification" is actually just the effects of reading a lot. Responding to one of the comments he says this:

I was quite the book worm as a kid but at some point I had to give that up due to lack of time. My dedicated book reading dropped way way down. I still read and wrote a lot thanks to my computer use but that wasn't the same as sitting down with a good book. With this technique I now have at least a couple of hours every day (more on the weekend when I workout) where I can do some genuine reading.

So he used to read a lot, but then read less (because: internet!) and now reads more again because it a) fits alongside other tasks and b) is faster due to speed-up.

Good for him I say, but the benefits he cites - better writing, better focus, general feeling of being smarter (better informed can feel this way) - sound like the benefits of reading to me.

[cromag]

Originally Posted by fjtorres

... Humans can process speech much faster than anybody can speak.
...

As a father who has observed (from a safe distance) the passage of two offspring through their teen age years, I'm not sure I buy that statement.

[fjtorres]

Originally Posted by cromag

As a father who has observed (from a safe distance) the passage of two offspring through their teen age years, I'm not sure I buy that statement.

Actually, I have an aunt like that.
Without a pitch control is was hard to get an exact reading on her top speed but she did approach chipmunk tones... Thing is, we're pretty sure she's an X-gene bearer, from other evidence.

[Katsunami]

People trained to do so can process TTS at tremendous rates. I've got an almost blind friend who has been using TTS for about as long as it exits. If he has an email read out to him, it takes about five to ten seconds, and sounds like "ablartedserupklasdafregjaklopigtibnarpladakal ..." to me. To him, it's a letter from his mother, half an A4 page long.

[lirazsiri]

Regarding pitch: I don't know about other TTS engines but with the Ivona engine which I am using on my phone I find that leaving it the default value works best. For what it's worth the Ivona engine speeds up without changing in pitch at all, so that might be the reason. It doesn't sound like a squirrel voice.

Regarding the scientific rigorousness of my anecdote: it has none but it doesn't claim to so thats kind of a moot point. This wasn't a scientific experiment in the sense that I have enough evidence to publish this as a paper in a Neuroscience journal. I'd need a statistically relevant sample size to even try to do that. Measuring IQ tests before and after wouldn't prove that this was caused by this in particular, by reading in general, by some other change I've made in my life, etc. It would show a correlation, but not a causation.

So at this point it's just this amazing subjective experience I'm hoping will encourage others to give this a try and see for themselves. If enough do, and the effect is as dramatic as I suspect it is then eventually the neuroscientists can look into this and perhaps prove and quantify the effect.

Again, I'm not trying to make the case that I have proven the effect exists myself. I strongly suspect it does based on my subjective experience and what I've since learned about neuroscience on how it might work but that's just a hypothesis and I might be fooling myself. I'm convinced but I understand why others that haven't experienced this for themselves would be skeptical because if the situation was reversed that's how I would feel.

The interesting thing is that this could definitely be tested scientifically. Especially if you go beyond statistical tools that can only show correlation, never prove causation, and actually look at what is happening in the brain to dramatically speed up language comprehension.

Note that tere is research to suggest that language plays a critical role in the development of intelligence though. Especially at an early age where small differences could eventually snowball into large differences in adulthood. Google "reading comprehension and intelligence".

[honneyStewPew]

Originally Posted by lirazsiri

I usually only lurk on the MobileRead forums but I signed up to share the interesting experience I've been having reading hundreds of books at gradually increasing text to speech rate. It turns out if you put yourself through a kind of training regimen you bump into the upper limit of speed supported by modern text to speech engines before you bump into the upper limit that your brain can be made to handle. Plus, there seems to be an unexpected intelligence amplification effect.

http://www.turnkeylinux.org/blog/int...text-to-speech

It's a bit counterintuitive but I'm hoping there might be a few eBook fans on this forum willing to give it a try and see for themselves.

Well...
that would explain my incredible intelligence.

[MikeB1972]

Well, I didn't have much in the way of audiobooks, but I gave it a go anyway.

I'm not sure if I feel any more intelligent but I now know Spot has a ball, a red ball.

Perhaps some more advanced listening would help.