[REL] sh4speech - voice command for SH4

[minsc_tdp]

Originally Posted by Digital_Trucker

I guess I'm the only person with the problem (or everyone else using the compiled version is running at 1024x768), but I can't get it to recognize my resolution. I tried making the change in the batch file and I've even gone so far as to attempt to recompile voice.pl with the resolution change (not having much luck there, gonna try some more today).

Any thoughts on why it wouldn't work by making the change in the batch file?

Whoops, sorry about that DT! That was a minor brain fart on my part in the code. I've updated it and posted it as 2.01.

[Digital_Trucker]

Originally Posted by minsc_tdp

Whoops, sorry about that DT! That was a minor brain fart on my part in the code. I've updated it and posted it as 2.01.

Hey, no problem dude! Thanks for the quick fix. As for the brain farts, my motto has always been "If you don't have an occasional brain fart, then you must not have a brain"

[minsc_tdp]

Originally Posted by Anti_Ship_Fella

well sounds great but i dont have a microphone

You know, I tested this on a laptop the other day that has a built-in mic. That mic sucks SO bad but yet it recognized great with this. You could probably pick up a simple desktop mic for less than $10.

Here you go, a whole bunch for less than $1

If you want to spend some money, I'm using this monster

[AVGWarhawk]

Install.........easy!
Voice recognition...........good!
Voice got confused response............sometimes.
Wife thinks I'm nuts talking to the computer......not uncommon for me to do so who cares?


Does the voice recognition get better over time? Reason I ask, I say bridge and it takes me there and then the binocs come up.

Other than that, a really neat program. Do not touch the keyboard but for very few things. The best part.....YOU FEEL LIKE YOU ARE SKIPPERING THE BOAT!!!! It is really cool. You say a command and you get a response of "Yes sir". Now, just to get it to recognize my command all the time and you got a winner here.

[minsc_tdp]

Originally Posted by AVGWarhawk

get it to recognize my command all the time and you got a winner here.

The speech recognition out of the box is decent but keep in mind, it knows nothing whatever about you and the idiosyncracies of your voice. It helps a lot to train it, and of course, to enunciate well, but mostly training.

It won't improve over time on it's own, as it has no way to know if it made a mistake.

Read the ReadMe sections on tuning your microphone and training the speech engine. Under Control Panel > Speech, if you go through a 10 minute training session, that greatly improves the recognition quality.

Also, if certain commands are being falsely recognized (like binoculars when you say bridge), sometimes it helps to change the voice commands slightly to make them a little more different. Perhaps "binoculars please" and "go to bridge", for example.

[Excalibur Bane]

Ah, ha! I got you this time! The list is updated before your sig is even updated! Beat THAT! Muhahaha! :rotfl:

[Digital_Trucker]

Originally Posted by Excalibur Bane

Ah, ha! I got you this time! The list is updated before your sig is even updated! Beat THAT! Muhahaha! :rotfl:

Gonna have to change your nick to "Speedy Gonzales" :rotfl:

[minsc_tdp]

Originally Posted by Excalibur Bane

Ah, ha! I got you this time! The list is updated before your sig is even updated! Beat THAT! Muhahaha! :rotfl:

LOL

one other thing everyone... in my zeal to keep sh4speech up to date for Trigger Maru, I didn't notice that TM really does change the layout of some of the command bar buttons, so a stock configuration is broken for a few commands now.

I've updated the system with two launch executables now, one for stock and one for TM, with separate sets of CSV files for each (though there's still just one Voice Command function - that won't matter, even though TM has a few extras that won't work in stock mode, no biggie.)

UPDATE: OK it is posted! with two executables, one for stock and one for Trigger Maru (but still requires stock key config in SH4, commands.cfg) Highly recommend everyone upgrade to this, as always just unzip into your current folder.

[AVGWarhawk]

Also, if certain commands are being falsely recognized (like binoculars when you say bridge), sometimes it helps to change the voice commands slightly to make them a little more different. Perhaps "binoculars please" and "go to bridge", for example.

Ok, so I can change a one word like "go to bridge" and this will help? The basic word of "bridge" still needs to be spoken yes? It is quite facinating that this works as well as it does. I hardly touch the keyboard. Also, when I fire up a game do I always need to say how many torpedoes and reset dials?

[minsc_tdp]

Originally Posted by AVGWarhawk

Ok, so I can change a one word like "go to bridge" and this will help? The basic word of "bridge" still needs to be spoken yes? It is quite facinating that this works as well as it does. I hardly touch the keyboard. Also, when I fire up a game do I always need to say how many torpedoes and reset dials?

So, here's what I mean. In voice_commands.csv, the third column is the spoken command (the second one is mostly ignored), so instead of:

4,Bridge,Bridge
57.5,Binoculars,Binoculars

You would make those:

4,Bridge,go to bridge
57.5,Binoculars,binoculars please

Then they would be sufficiently different to prevent overlapping recognitions.

Also do the Configure Microphone under Control Panel > Speech. This only takes moments and can help a lot. Make sure when you speak the test phrases that your level meter sits right in the middle of the green area while speaking. If not, adjust your microphone gain/volume levels accordingly. This guarantees the system hears what you're saying at normal volumes - too quiet is hard to recognize, and too loud causes weird distortions at the top-end that sound staticy and hurt recognition.

You do need to say how many tubes once each time upon loading sh4speech (you can do it before the game loads if you want). I suppose I could cache this in a file somewhere so you have to say it less often, but for now yeah you have to say it each time. Read the section in the Readme called STARTING A GAME for details on how this works, and the other two similar commands, and when exactly you need to do it.

I'm glad to see you're getting great results. I've been trying to convince the skeptical that the voice recognition is excellent - Microsoft and various universities have been working on this stuff for many, many years, and we get to reap the benefit of their efforts. I think this sub genre is a perfect application of it, since there's not very many commands (the fewer commands, the better the recognition), and there's not a lot of similarity between them (bridge and binoculars aside). But I think the number of commands is the key thing.

Picture this: If I said I'm going to say one of two words - "bridge" or "torpedoes", you could probably pick which one I was saying from 100 feet away, underwater with my head in a paper bag. The opposite of that, if I was going to pick any word in the english language, conditions would have to be a lot better for you to figure out what I said. The same applies to this system. A lot of speech recognition stuff is used for dictating into MS Word and for that, it's trying to use the entire dictionary, so it screws up a lot.

[minsc_tdp]

Hey all how about this:

"captain on deck!" or
"captain on the bridge" = start recognizing commands
"XO you have the bridge" = ignore commands until you return to deck

Man that's dorky... cool but dorky as all hell.

Just posted an update 2.04 which fixes the fuel/battery/o2/co2 buttons when using the command bar in unlocked/autohide mode.

[Digital_Trucker]

Originally Posted by minsc_tdp

Hey all how about this:

"captain on deck!" or
"captain on the bridge" = start recognizing commands
"XO you have the bridge" = ignore commands until you return to deck

I like the idea of a listen/don't listen command (with the facility to change the spoken command). I am one of those unfortunate people with no mute button on my headset and lots of screaming/screeching people running around in the room. It really gets aggravating when I'm setting up a shot with my last fish and my son's girlfriend comes in the room all dramaed up about something and fires my last torpedo at the sun.:rotfl:

[panthercules]

Yep - sounds like another good idea - I'd suggest another couple of command options:

"X O you have the conn" (sounds better than "bridge" when you're submerged)

"X O you are relieved" (for starting listening back up)

BTW - I'm still testing and tweaking, but so far this is looking really good and lots of fun. One question though - have you changed any of these .csv files since version 2.0b?:

voice_commands.csv
key_commands.csv
key_codes.csv

These are the ones I've been playing with in my customization, and I have them merged based on 2.0b - just want to make sure I haven't missed anything in those files in the newer versions (gonna go look for that comparison program you mentioned a while back, but haven't got it yet).

Thanks for all the great (and FAST!) work

[minsc_tdp]

have you changed any of these .csv files since version 2.0b?:

voice_commands.csv
key_commands.csv
key_codes.csv

Here's a comparison report versus 2.0 using the FC command line (which you also have) (sorry I deleted 2.0b) As you can see all I did was add a few commands:

Comparing files voice_commands.csv and VOICE_COMMANDS20B.CSV
***** voice_commands.csv
516,Patrol Course,Set up a patrol course,
516,Patrol Course,search pattern,
517,Time To End Course,Time to end course,
***** VOICE_COMMANDS20B.CSV
516,Patrol Course,Set up a patrol course,
517,Time To End Course,Time to end course,
*****

***** voice_commands.csv
517.5,Range To End Course,Range to course end,Trigger Maru
517.5,Range To End Course,maximum range,Trigger Maru
517.5,Range To End Course,maximum range at current speed,Trigger Maru
524,Recognition Manual,Recognition manual,
***** VOICE_COMMANDS20B.CSV
517.5,Range To End Course,Range to course end,Trigger Maru
524,Recognition Manual,Recognition manual,
*****

***** voice_commands.csv
557,Send Bearing To TDC,send bearing to TDC,
557,Send Bearing To TDC,sonar send bearing to TDC,
558,Send Range To TDC,send range to TDC,
558,Send Range To TDC,sonar send range to TDC,
558.1,Follow Nearest Warship Contact,follow nearest warship contact,Trigger Maru
***** VOICE_COMMANDS20B.CSV
557,Send Bearing To TDC,send bearing to TDC,
558,Send Range To TDC,send range to TDC,
558.1,Follow Nearest Warship Contact,follow nearest warship contact,Trigger Maru
*****

[Rockin Robbins]

Trigger Maru, and possibly other mods, change the commands.cfg file. The commands.cfg file can be restored from the !BACKUP stored by JGSME to fix sh4speech, I don't know if that will have any effect on Trigger Maru or any other mods that have custom commands.cfg files. Of course you could go and change all the key commands in the CSV as well but this is really designed for the default key configuration. Trigger Maru has some additional command bar items that are supported by sh4speech.

That's a deal killer for me. I'm not going to settle for "I don't know if that will have any effect on Trigger Maru" when you already have a different batch file for Trigger Maru. Either this thing is compatible with Trigger Maru or it isn't. Looks like it isn't without some hacking to make it talk to Trigger's commands.cfg file. Ducimus wouldn't have changed it unless there was an advantage or a necessity to it. If you went through the process of making a different installation for Trigger Maru why didn't you do the whole job and make it work properly?

If I go through the csv file and make it actually work with Trigger Maru, will you use it? Trigger Maru is the mountain, you are Mohammed. You must go to the mountain. The mountain will not come to you.