Just a thought, I was reading an article on speeding up recordings of lectures (NYT link) and I thought of the pitch bend feature of V3. Would it be possible to build from this feature to listen to recordings at 1.5x or 2x the speed, say for audiobooks, only without bending the pitch? I know that the techniques used in the article involve cutting the recording when a constant sound exists (like for long vowels) but could this somehow be adapted to the Empeg?

I dunno about the techniques mentioned in the article, but there are well-known algorithms for speeding up/slowing down music without affecting pitch, but it requires some pretty heavy processing power and the quality tends to degrade if you go beyond a certain amount. My guess is that for spoken word the same algorithms would work. Whether the empeg would be powerful enough to chug through such a task is doubtful though. The software I used couldn't even do it in real time on a Pentium II class machine.

There are plenty of algorithms out there which allow you to change the speed of the recording without changing the pitch. Some of them are very fancy, and they have to do a lot of math (calculus, actually, IIRC) to make it sound natural instead of electronic-y.

The best versions of these algorithms are used in professional audio editing packages, and the details of each are kept trade-secret because it's a selling point for some of these packages.

Note that the converse is also something that is done often in these packages: Changing the pitch without changing the speed. Same principle, similar algorithmic challenges, just along the other axis.

When the pro packages do this, it's not usually in real-time, it requires that they churn pretty hard on the file to produce a decent result. And you can only go so far off of the root pitch or time before it starts to sound strange.

I agree, this would be good for audio books. Not sure if the empeg's got the kind of CPU to do this well, but perhaps a lower-quality version would be a nice feature.

Hrm.. it would be great to have this on-board the Empeg, but the next best thing I guess would be to reprocess the MP3 or WAV beforehand.

Hmm. Back when I played around with this stuff I had tried out Prosoniq Time Factory which ain't cheap at $449. But they now appear to have a stripped-down version, EZ Time Stretch that will change time but not pitch for $179 (Time Factory would let you do both time and pitch if you wanted.) Back then there was a demo available for Time Factory, but no demos of either product seem to be available (though they do have sound samples to demo Time Factory.)

Maybe you should sneak over to KazAA or the like and, copyright morals permitting, give EZ Time Stretch a shot, and buy it if you like what it does.

Edit: Looking closer at the page, it appears they no longer sell EZ Time Stretch, so it sounds like the P2P route is the only way for that partiuclar program.

I'm going way out on a limb here, as I have no idea how the mp3 codec really works.

However, I do see a vague similarity between this idea and VolAdj (which, if memory serves, once upon a time doubters said couldn't be done in the empeg - thank you Richard, et al).

Would it be possible for an algorithm to look so many frames forward in the stream, compare them, and if there is a mostly static waveform, drop some of the frames? I would think that would shorten longer vowels and blank spaces.

My 8 year old digital answering machine does this with blank spaces between words; how hard can it be? (ducking and running )

-jk

Would it be possible for an algorithm to look so many frames forward in the stream, compare them, and if there is a mostly static waveform, drop some of the frames?

Dare you to try that and see how it sounds.

Those calculus-heavy algorithms I mentioned above aren't there because the people who wrote them were trying to do it the hard way. They're there because it truly is difficult to do it intelligibly, and there aren't really any short-cuts.

I don't want to hijack the thread, but does anyone have insight regarding what current devices (ie: handheld PC, MP3 recorder) work well for recording lectures, then accelerating them during playback (either on device, or on a PC)?

My daughter is trying to use a compact tape recorder at university, but it isn't really sufficient for the task.

Thanks in advance.

I agree, this would be good for audio books

Ummm... why? Do you use Pitch Bend when you listen to "Moving Pictures"?

The object of listening to an audiobook isn't to get through it as quickly as possible. Just skim through the Cliff notes if that's all you want.

Here -- I'll save you 12 hours and 21 minutes: A hobbit finds a ring that is really powerful and if the bad guys get it it'll ruin the world but he throws it into the cracks of doom where it is destroyed and everyone lives happily ever after.

I'll take my audiobooks at normal speed, thank you, where I can really enjoy them.

tanstaafl.

Maybe we are talking about informational audiobooks, not entertainment audiobooks. The article I linked to had students listening to lectures on CD and DVD. I think if you could get used to it, NPR could possibly be listened to this way. I never was able to listen to all of the NPR news broadcast when I used to commute, but it could be streamed from the web, stored, compressed, etc.

Hey, whaddya know. (If you don't like registering at NY Times, use foobar123 for the user name and password).

In reply to:

---

but he throws it into the cracks of doom where it is destroyed and everyone lives happily ever after.

---

HEY HEY ... Easy on the spoilers

\me throwing away my cinema-preorder of The Return of the King....



cheers, Thomas

speaking of which, is there any new information about that trilogy marathon that they're supposed to have the thrusday night of the release? lotr1:se, lotr2:se, and when lotr2 ends, it becomes midnight and lotr3 starts playing. actually, now that i think of it, there better be intermissions.

Yeah...they posted the list of theaters that will be doing it on the official website. It happens on Thurs. Dec. 16.