[Feedback][P][✓] Faster text appearing

[Sligneris]

That is something that is bugging me a lot. First, I think that text appears too slow. Also, blip sound could be a bit... softer? Something like that.

Check out how it looks here.

And now go, and play something on AAO. There's definitely a difference here.

I realize I can be a bit nitpicky, but I wonder what are other's thought on this.

[Kroki]

It's right that on the editor the blip sounds is louder, when you raise the volume to hear the music it can be a bit disturbing. Also very right that the text is slow. u_u

[E.D.Revolution]

Text speed is still too slow about two years (or maybe even longer) after the issue was raised. Used to get around the problem by going with a 2.0 but now that threshold is gone lower. What the means is that 2.0 is too fast for normal use. I personally use 1.25.

Sounds seem to be a problem in general. The ones here and the ones compiled on CR don't always match.

[Meph]

The problem is that the speed of your web browser has a big effect on the typing speed. I'm not even sure if anyone can see the typing speed that Unas intended.

[Unas]

Well, Meph, if it increasing the speed in the editor has a noticeable effect, it means that their browser is actually capable of performing the typing faster, so it probably isn't a technical limitation anymore.

Now, if someone wants to help on that matter so I can fix it in V6, this is what you could do :
- Find a frame that contains a long text, typed continuously at default speed, in the original games, and measure the time between the apparition of the first and last characters.
- In the V6 editor, create a frame with exactly the same text contents.
- In the V6 player (in Chrome if possible), play this frame and measure the time between the apparition of the first and last characters, in the same way as before.

Give me these two durations, and if possible a link to the V6 trial where you tested that frame.
Ideally, a video of the frame from the original frame being read would be good, too.

With that I can update the default typing speed of V6, and update the V5 to V6 conversion scripts to take that new factor into account.


As for the sound, in fact we should probably harmonise the volume for all default music and sounds anyway. Ideally, if someone had time for this (listening to all default music and sounds, and use tools like Audacity to reduce the sound a little but on loud files so that it fits better with the others), it'd be great.
Without going so far though, I can just reduce the volume of the bleeps, for starters...

[GanonZD]

I'm on it.

Here's my report:

Introduction

In fact, typing speed is one of the problems that I have been worrying about myself many times; the default text speed seems awfully slow, and I have been very close to making a post with a similar suggestion many times before. But what has stopped me every time is this very question: Exactly what is the "default" speed? At first, the answer seems to be obvious: It would be the speed that occurs most frequently; the one that characters use when talking "normally", i.e. without too much enthusiasm, not too energetic, affected, or eager. That means that frames that contain shakes, flashes, sound effects or similar are to be ignored. But when looking at actual AA footage, what strikes me is this:

• First of all, AA uses quite a lot of speed levels, and it is not particularly easy to decide exactly which one is the most "default" one.
• Long text frames without any shakes, flashes, breaks because of punctuation, or change of speeds are surprisingly rare; these effects are in fact used much more frequently used in real AA games than in a typical AAO game. Finding a long frame with none of these frames is not entirely easy.

So, because "default text speed" appears to be a relative matter, certain choices have to be made in order to even make the examination that you request.

My approach

So what should be the approach to defining something like a "default text speed" for the entire series? I decided upon this choice:

• First of all, I chose that the sequence investigated should be from AA4, since I think that --- at least technically --- this is where the main AA series has reached its most "final" form. So this --- rather than any solutions that have been uses in the previous installments --- should serve as the bases.
• Secondly, I decided to pick a sequence from one of the initial cases rather than the later and more "dramatic" ones. Though I don't think it is so, I cannot eliminate the possibility that the "normal" typing speed may increase a bit in the final cases of the games.
• Finally, I decided that the sequence in question should be part of a cross-examination, since these are generally formulated in a very neutral, unaffected, and matter-of-fact way. I believe that if anything is the "default" AA language, it is the language used here.

After looking through case 4-1, I decided upon this sequence:


The sequence occurs at this point. As seen, its exact content is:

The girl was knocked out cold, and Smith was uncontrollable. I left to call the police.

Feel free to count yourself, but I count 87 characters including punctuation and spaces. I have recreated the sequence in V6 here.

Analysis

The analysis was done by downloading the previously linked recording from YouTube and editing it using VideoPad from NCH software, where the sequence in question was cut out and its exact time measured. I decided to cut it so that the first video frame was the very first one that included text from this sequence, and the last frame was the very first that contained the final dot. Please note that he first of these video frames did contain text, namely the first "T". Thus the text box was never empty; the new text started appearing at the very moment the old text disappeared from the box. Through close analysis, I found that the sequence appeared to play completely continuously, even at the commas and the full stop. Therefore, I conclude that the example is representative.

The results I found were the following:
Original AA game:
Length of clip: 2.8 s
Number of characters: 87.

Average time per character: 2.8/87 = 0.0322 seconds per character.
Average characters per time: 87/2.8 = 31.0 characters per second.

Recording the sequence in V6 using the screen recording device CamStudio yielded the following results:

V6 version:
Lenth of clip: 4.3 s
Number of characters: 87

Average time per character: 4.3/87 = 0.0494 seconds per character
Average characters per time: 87/4.3 = 20.2 characters per second.

Well, like in any good report, we also need a section with the

Sources of error

First of all, the choices I made could have been wrong, and perhaps the speed in that particular CE is not the default speed, if such one exists. It would probably have been better to make more measurements of more sequences, but I do not have time or motivation to go through this entire process again, but I do encourage others to "reproduce the experiment".

Second of all, the video clip I used was merely a recording from YouTube made by someone else, presumably using an emulator and a recording device. This is far from a perfect solution, and I cannot guarantee for the quality of that recording, and for whether or not it played at exactly the original speed, though it seemed to be in accordance with my experience with playing AA on DS devices. And anyway, the text speed seemed perfectly acceptable to me.

Also, I am not entirely sure if the speed of Ace Attorney (and DS games in general) varies between 50 Hz and 60 Hz regions. And we definitely cannot know at what exact speed this emulation ran. But again, the recording seemed in accordance with what I find to be "good" Ace Attorney text speed.

Finally, there is the quality of the software I used myself. I merely googled for the best freeware editing program available and found VideoPad. I then used CamStudio to record it on-screen. Again, I cannot guarantee for the quality of these products. Particularly, the sound quality of the CamStudio recording was horrible, though the video quality seemed sufficiently good. When playing the video file, it seemed to be quite close to how AAO rendered it.

Finally, when playing the AAO version, I notice that the beep sounds make a small break when the word "call" changes line. This may be some kind of bug, but when playing the video in slow motion, it did not seem to affect the actual typing speed. Therefore, I conclude again that the example is representative of the function of AAO.

Generally, I conclude that the measurements I made are precise enough to be used as a guide for an adjustment of the default typing speed. But again, I highly encourage others to make similar analyses to other sequences from the official games.

Conclusion

I suggest that you increase the default typing speed by a factor 1.5. This is very close to the relation between the numbers I measured, and by using this relatively round, "nice" number, the relation between the new and the old typing speed is relatively simple. Furthermore, such a multiplication will not involve much rounding up or down, so the conversing process is easily revertible and can be reverted if the result should not be in accordance with our expectations. I have created this V6 example to show how the same text would look with a text speed of 1.5. It seems pretty close to the original game to me . Furthermore, applying the exact same method for measuring as before, I found that this sequence lasted exactly 2.8 seconds!

One final suggestion...
Now there has been lot of talk about the text speed and the beep sound quality. But why not combine these two issues into one?

The beep speed also needs to be adjusted, since in the official games, it varies along with the text speed. If web browsers do not contain features that can change the beep speed in the browser itself, I imagine maybe having three to five different sets of beep sound files to be played at different text speeds. This would of course only serve to approximate the typing sounds, but I think the solution is accurate enough.

Finally, one important piece of information that I almost forgot: It should be noted that I am playing everything in Chrome on a three-year-old laptop with Windows 7. However, my PC works sufficiently fast IMO.

Hope this lived up to your expectations, Unas.

[Unas]

xD
I wasn't expecting such a detailed dissertation, but thanks, there's all I need here ^^

You're wrong on one point though : if I use your factor of 1.5, it means I have to divide the speed factor for each frame by 1.5, which in most cases requires rounding (1 / 1.5 = 0.6666666666....), which makes conversion not exactly revertible. But I don't mind at all, since I don't intend to change the default speed every now and then ^^
I've implemented this default speed change in my working copy, it will show up next time I update the online beta

[GanonZD]

Well, including all of these details was pretty much an attempt to say "This is my conclusion, but it's based on these factors, and it might not be entirely accurate, so take it with a grain of salt unless somebody else decides to reproduce and confirm the result."

[Unas]

I understand, and I like it that way, though it might be a slight overkill

Also, about using different sound files for different speeds, it's a possibility and a good idea, but it's neither an instant fix nor an urgent one at all so I won't bother with it right now ^^

[Meph]

Using different bleep speeds seems like a good compromise. But I still think it's worth trying to create each bleep manually. Would it be too laggy to do it like the following?

1. Create an MP3 file containing just one bleep
2. When the player arrives at a new frame in the trial, measure the frequency for three characters to be typed out in the dialogue box.
3. Play a bleep at that frequency until you reach the end of the dialogue?

[GanonZD]

Well, I'm more worried about the restarting process. Don't we risk some kind of delay?

I understand, and I like it that way, though it might be a slight overkill

Well, if we're about to make a significant change that affects the speed at which every single new trial plays by default, meaning that it affects the entire playing experience quite a lot, then I would require some certainty in your place . Preferably there should be results be several people who measure independently. So that's why I decided that it could not be done precisely enough. It's not my ordinary free time hobby to sit down for several hours to measure and double check and then write four pages long reports . It's because I concluded that what I dealt with could have consequences if done the wrong way.

[Bad Player]

Getting the textspeed up would be nice, but I don't think we need to worry 100% about the blips. I think it'd be nicer to get the blips better synced with the start/stop of the text than to get it 100% like the games.

[Unas]

I should point out that the update I pushed yesterday evening contained the fix for this.
Default speed in V6 trials now faster; however, imported trials from V5 will remain at their usual speed since the speed factor is converted for better backwards compatibility.

For your information, I used Ganon's numbers :
Speed 1 in V6 corresponds to speed 1.5 in V5
Speed 1 in V5 corresponds and is converted to speed 0.67 in V6