iGo Primo TTS Swallowing First Part

Hey,
So I got my new Head Unit which is just a China 2Din. I've had a couple of these but I've never had this bug. Apparently, I'm not the only one though.

With TTS activated, it seems to swallow the first word of announcements. Any idea why?

e.g. When it gives a route summary it usually says

"...y A34, A50, M1" etc etc rather than "Route Summary A34, A50, M1"

It does this on instructions to, it sometimes says "300 Yards, Turn Left" instead of "In 300 Yards, Turn Left". It's not an isolated incident and it's something to do with iGo, but I just don't know what settings need to be changed. I've thrown together a few sys.txt files to see if I can fix it, including one with loads of memory, cache, and TTS memory reserved. I hear that deleting the sys file altogether can help, but I need HTTP access.

Any advice would be smashing :) Surely someone else has had this problem :D

Non-TTS voices are fine, so there is always that :)

Serena seems to do this a lot, Daniel is slightly better, I don't have Kate, so maybe a different TTS voice would help fix it?

It's possible that you got bad incomplete download of voices. If you want to try Loquendo voices instead of NUA post here.

Where would I get them? I'm fairly sure they aren't incomplete because I just copied them from the old SD card lol but I'll try anything right about now :P

Everything you need for Loquendo is in the blue box http://www.navitotal.com/showthread.php?912-TTS-voices-and-components Download and copy/paste the LOQ_TTS folder to the root of your Primo folder. Copy/paste the 4 .dlls to the root of the Primo folder after you download them. Download and copy the English voices [all the files] to content\voice. Make sure you read the necessary edits in the blue box to include also.

OR

There is this http://www.navitotal.com/showthread.php?12708-Variety-of-Primo-voices

Thanks, will give them a go. Ultimately, if I don't get it working, it's not the end of the world. Despite that little bug I prefer this Head Unit much more to my previous one, which Ill sell for easily what I paid for it.

I have a soft spot for China Navs and China Head Units, for me, they are clear proof of why brand culture sucks!

No luck with the Loquendo voices. Shame. It looks like it might be a sort of bug in both my head unit and iGo Primo.

In the head unit settings, there is a function to disable Music/Radio etc while the Navigation is on, so all you hear is the Instructions from the Sat Nav. With this function turned on, TTS is perfect. There are no dropped words or letters and the whole thing seems to work beautifully smooth. So you'd think surely it's the head unit? Well no, it's not quite that clear. If you disable this function but then turn Natural Voice instead of TTS then it works perfectly. So it's like there is a bug in iGo + a bug in the head unit causing the thing to go wrong.

If the bug was affecting natural voices as well, I'd have said it's the devices sound mixer reacting slowly to the wince sound output, ie, turning on the sound channel when it detects sound output, all I can assume is the extra processing TTS takes just before it speaks is interfering with the mixer causing the delay.

Wince could possibly be opening a serial port to the sound mixer when sound is detected, and the TTS processing slows this process down?

Fatboyfun wrote:If the bug was affecting natural voices as well, I'd have said it's the devices sound mixer reacting slowly to the wince sound output, ie, turning on the sound channel when it detects sound output, all I can assume is the extra processing TTS takes just before it speaks is interfering with the mixer causing the delay.

Wince could possibly be opening a serial port to the sound mixer when sound is detected, and the TTS processing slows this process down?

Well, I figured this too. So I activated 'Dings' before instructions, but it doesn't seem to have made much difference.

Is there anyway to add a half second delay in front of TTS instructions? Surely that would give it enough time to sync up?

What is in the sound section of the sys.txt?

Currently this:


[sound]
sound_q_length=250
use_method=1
csv_enabled=1
;queue_length="200"
ding="1"
mute_os_key="1"
wheelsound="1"
pause_mp3_sooner="1"

If there is a way of adding a delay maybe? I need to cut down my sys.txt a bit. It's one I made with that SysConfigurator thing and it's added a load of crap that I'm sure doesn't need to be there!!

Ive seen this on a couple of in car gps systems Ive fitted for people, even still on my wifes one.
Ive posted before and played with the sys.txt with no fix. Mainly on TTS phrases.
Ive been meaning to play with the sound file.
"Red light camera" becomes "... light camera" like you say.
What I am thinking of trying is to add an extra word to all the phrases. ie "[color=#FF0000] A[/color] Red light camera" so when it drops the 1st word it wont matter.
If you do it or before me or find another fix please post back.

Bazzle

I've seen this sorta work in other things but it might just help here also. In the TTS voice file, there is a file called "info.ini. Open that file and look for tts_rate=some number. Play with that number and see if it helps. Remember your old number just in case.

bazzle wrote:Ive seen this on a couple of in car gps systems Ive fitted for people, even still on my wifes one.
Ive posted before and played with the sys.txt with no fix. Mainly on TTS phrases.
Ive been meaning to play with the sound file.
"Red light camera" becomes "... light camera" like you say.
What I am thinking of trying is to add an extra word to all the phrases. ie "[color="#FF0000"] A[/color] Red light camera" so when it drops the 1st word it wont matter.
If you do it or before me or find another fix please post back.

Bazzle

This is one I've heard on other forums. I think the file you have to edit is config_database.lua in the voice zip file, although I heard there was a way of overriding it in the sys.txt file, but no idea how. I read about it somewhere and it seemed like a great idea but I can't find the link now! This was the whole point of activating the Ding setting, but the problem is the voice is output too long after the initial Ding, otherwise it would work perfectly. I was thinking most instructions output by TTS are pretty easy to add something to, for example when it says : "Turn left at the roundabout, taking the first exit"
You could easily add a Now at the start so the sentence reads: "Now turn left at the roundabout, taking the first exit".

The initial word would take most of the delay out, but not be confusing if it was output as well.

chas521 wrote:I've seen this sorta work in other things but it might just help here also. In the TTS voice file, there is a file called "info.ini. Open that file and look for tts_rate=some number. Play with that number and see if it helps. Remember your old number just in case.

I'll try this tomorrow. I think they're 100 by default right?

I was thinking about this further and perhaps we've got this all wrong after all... On most forums, the suggestion is to switch the TTS priority to Very High or High on the CPU side, so it's processed as fast as possible and then passed to the head unit output. Think about it... the problem is that there is a lagg of some sort between WinCE and the bespoke implementation of the audio hardware, probably a delay in the mixer adjusting the output to the unit. When WinCE outputs any kind of audio, there is some sort of interrupt and the lagg is occurring due to a sound mixer delay on the WinCE hardware, probably caused by a delay in the threading.
So... surely the best thing to do is the exact opposite of high prioritization? By prioritizing TTS highly, the vocal audio is compiled quickly and output before the mixer settings have been adjusted and the interrupt is sent to the hardware. If we change the TTS priority to low or very low, the likelihood is that the compilation time will be increased, maybe not by much, but the mixer will have a set system priority so the interrupt will be sent at exactly the same time.

I will test this theory tomorrow, it may just cause the TTS audio to break up, but I imagine it won't because TTS is not about the output, but the processing of text to speech.

Whatever works. Please post your results and what you exactly did to be successful. Others might have the same problem. Thanks.

chas521 wrote:Whatever works. Please post your results and what you exactly did to be successful. Others might have the same problem. Thanks.

Others do have the same problem. It's quite a widely reported issue. Haven't tried yet, will let you know when I have.