I just installed Pilot2ATC and am using the Amazon Polly voices…holy smokes what a game changer. Wish I would have had this a long time ago.
I was looking for more natural sounding voices and came across this thread. Thanks for the tip!
I have the same issue that sometimes the Amazon Polly voices work in P2ATC, but sometimes they just suddenly stop for no discernible reason.
I am trying to use the Neural versions of the voices, and they work sometimes. Any reason why we should stick to the Standard ones aside from the price?
I thought I had read somewhere that the Neural voices will not consistently work with P2A. I stick to only standard voices and they typically work well.
Completely the same here.
Awesome thanks for the tip!
I initially thought Neural should sound better, but surprisingly the Standard ones actually sound more natural to me. The ones I like the most are the Conversational versions, but those only exist for Joanna and Matthew.
Also, for some reason having random voices seem a lot more unstable and prone to voices failing to load. I’ve removed all random voices and set them as fixed like so:
Flights have been pretty consistent so far. The voices are leaps and bounds better than the default MS ones. Has anyone been using this long enough to see how much it actually costs per month?
The only other irritating thing is they say “whynds” (long ‘i’) instead of winds. I think this is because Pilot2ATC has the phraseology saying “whinds” with an ‘h’. I tried creating a Lexicon alias to work around this, but unfortunately Lexicons don’t seem like they work with the Windows plugin. I sent an email to Dave and hope those “whinds” are something he can fix on Pilot2ATC’s end.
Good findings, thanks. I will give Polly another try with your advice.
I have definitely noticed that it is much more stable as well when you have the different voices assigned. I have the ATIS assigned to a default voice (Microsoft David I believe). It sounds great as from what I have heard, most ATIS reports come from a computerized voice.
Thanks again for sharing!
I suspect the ‘h’ in ‘whinds’ is there to differentiate the pronunciation between two versions of wind - “Storm force wind” (to rhyme with tinned) and “Wind up the clock” (to rhyme with dined).
As far as cost goes, Amazon Polly is essentially free for the first year - allowing 5 million characters per month.
Can someone do an idiots guide how to setup the amazon polly voices to work with p2atc, ive created an account got the windows plug in which says pick a voice which is blank. I went to the speech recognition settings to change to polly voices as instructed but I only have the default windows 10 voices available in the drop down. I’m lost
Have you set up your Identity and Access Management (IAM) within your Amazon Web Services (AWS) account? Unfortunately, AWS can be a bit of a steep learning curve and I don’t know of any step-by-step guides for setting up Polly voices.
To make setting up even more of a pain, there is also an issue with P2ATC and Polly voices that often prevents the voices working - and the only option (at least as far as I have found) is to restart P2ATC. There does not seem to be any pattern to this behaviour and it can can often take 3 or 4 attempts at starting P2ATC before the voices work. Fortunately, once they are working, they stay working - I simply test a Polly voice immediately after starting P2ATC and restart it if it does not work. After this, the voices work pretty flawlessly.
Exact same for me. I would recommend also only using the Polly “standard” voices.
Thanks for replies, I got it working and yes it’s a right cluster setting it up. Switching pta off and on sorted it eventually. A big thanks to Clumsy on YouTube for pointing me in the right direction.
I found the problem occurred just as regularly with standard voices as it did with neural or conversational?
I find neural voices better than standard but the conversational voices are really a huge step up in terms of fluidity and intonation, so I use one of those exclusively for my co-pilot. For everything else I use the remaining conversational voice and mainly neural - usually randomly assigned.
I’ve managed to set up the polly voices with my pilot2atc. I certainly found it a bit of a challenge but got there in the end.
Anyway, they’re brilliant and work flawlessly the majority of the time.
However I occasionally get a popup error message that I think is some kind of connection error which I then have to acknowlege. This is rather an immersion spoiler and ok if it does drop the odd messge it’s not the end of the world and it would be preferable to the popup.
Has anyone else experienced these popups themselves and more importantly figured out how tho disable them.
I do have a 100 Mbit broadband connection that seems adequate for everything else.
Any kind of help would be appreciated. Thanks…
So good news: I got in touch with the aws rep for one of my employers, explained there were troubles using Polly within windows applications, using p2atc as an example together with a development reliability question, and they escalated it up to their polly devs.
Hopefully we can see some reliability improvements in 2022!
Such a game changer. Obviously, you can tell though that it is definitely not made for the general user as setting everything up through Amazon complex console is a real pain in the a**. You have to create at least two different types of account, different passwords, different generated keys to handle and etc, prompt DOS-like command lines to handle, and etc… you can tell it is definitely catered for the developer crowd, not the end-used crowd obviously.
Which raise a bit of a frightening caution flag regarding prices since it’s obviously made for professional use first and foremost:
Supposedly, if you stay to standard voice(s), after a year, one can assume a regular usage of P2ATC shouldn’t go above the million of characters, which should only cost you 4$/month. However, it is not absolutely clear whether it is for a single standard voice or if costs will pile up if you use different voices but I suppose not (other than the standard vs neural thing).
One thing annoying too is that I haven’t found any sort of counter managing your Polly usage in the console. Let’s pray we won’t end up having awful surprises in the end $$$…
I ditched Polly about a month or so ago after testing our Cereproc. They work 100% of the time and overall sound better than Polly.
Was running a mix for a bit (6 Cereproc and then Polly) but again having to constantly restart to get Polly to work was a bit annoying. Bit the bullet and bought another 7 Cereproc voices.
13 is a good mix, they all sound great and work 100% of the time.
Each Cereproc voice cost 26£ according to their website. Maybe there are bundles? Cause if not, it means:
13 x 26 = 338£ which is roughly 457$ / 400€ for 13 voices.
That’s quite the investment for voices that I found even more synthetic sounding than Polly.
That’s approximately 9 years worth of Polly voices… quite the bad deal IMO, at least for a majority of users with tighten budget.
When you’re using them side by side in P2ATC, they sound way better than Polly.
If you are interested in them, e-mail Cereproc directly and work with them on pricing. I won’t tell you exactly what I paid, but will tell you it was less than 50%.
I’ve ditched the Polly voices as they were frustratingly unreliable. I’ve now got four voices which I alternate. Deskshare sell Ivona voices for £22 (30 bucks) which is a bit cheaper than cereproc and they work well.
PorcoRosse, there is a page in your AWS account where it shows what current usage and cost is with your Polly voices. I can’t explain how to get to it but I do know it’s there as I was clicking around one day.