TTS invocation provides no initial audible feedback about pure-maps HOT 15 CLOSED

I have worked on TTS a bit and made a set of standard messages (starting, rerouting, ...) which should play on corresponding action (unless a newer message comes meanwhile). Please reopen if it doesn't work in the next release

from pure-maps.

P.S.: The output of Pure / WhoGo / Poor Maps WRT this issue, when started at the command line has not revealed much in the past, beyond pico2wav was called, when TTS works, and not, if not.
Furthermore this issue has also been reported by users of Mimic on TMO.

If you have any suggestions for more comprehensive logs or if you think it makes sense to capture the output of a recent Pure Maps, when started at the command line, please let me know.

from pure-maps.

I also had the impression that sometimes TTS didn't work in WhoGo Maps and in earlier versions of Pure Maps. But in the last few updates of Pure Maps the issue hasn't occured again. I mostly use Stadia Maps as router, sometimes OSM Scout, too.
The only thing I've noticed in every version so far is that directly after starting a freshly calculated route there's no TTS for the "initial" directive, same after rerouting. Only from the next one on directives are spoken, for this to happen can take a long time when starting navigation while driving on a long road going straight, without turns or roundabouts. Maybe that's what made my impression I've mentioned above.

Pure Maps 1.7.0 (PicoTTS) on Jolla1 Phone

from pure-maps.

Triggered by @Pohli's comment, I retested this with PureMaps 1.9.0 using PicoTTS 17.08.10, OsmScout as routing library provided by OsmScoutServer 1.12.2 on a Jolla 1 under SFOS 2.2.1.18 (the original report here was for PureMaps 1.3.0, while @Pohli tested with v1.7.0):

Yes, TTS navigation has become even more stable compared to early releases of PureMaps (and much, much better and reproducible compared to PoorMaps and WhogoMaps)!

I fully support his observation, that currently the only reproducible flaw left is:

The only thing I've noticed in every version so far is that directly after starting a freshly calculated route there's no TTS for the "initial" directive, same after rerouting.

But this happens in only ca. 50% of my tries with PureMaps 1.9.0, about half of the tries it works immediately (e.g. when hitting "Reroute" repeatedly).

And as before: When PureMaps once started "talking", it always continues to work fine until exiting the navigation mode.

P.S.:
@rinigus, I asked myself, if PicoTTS could be easily updated to a newer version (even though I do not expect that to cure this issue), only to find out, that it originates from Google's AOSP tree (as you mention).

Maybe the effort by @gmorenz, providing scripts for fetching and compiling PicoTTS under a regular Linux distribution is helpful for you (as I assume you basically did the same, but manually). As PicoTTS seems to have been declared EOL by Google (so maintenance has more or less stopped), it may make sense to create a presumably "final" release for SFOS from it.

From my experience, PicoTTS is by far the best working TTS for SFOS:

• Mimic is "English only", really resource hungry and I never had it working in Pure/WhoGo/PoorMaps (but at the command line with garbled output).
• I never found a Flite compiled for SFOS, although it is mentioned as a TTS engine.
• I also haven't found a eSpeak compiled for SFOS (although it is also mentioned as a TTS engine), but having tried eSpeak as an Android TTS provider under AlienDalvik, I can can tell that eSpeak sounds funny, but the TTS quality is simply unbearable, rendering it unusable!

from pure-maps.

Re TTS engines:

PicoTTS is in this sense "final" compilation. There is no further development going on with it and I think I did use the "final" version.

Mimic and Flite are the same, except Mimic has AP voice. Try to use female voice model, that I would expect to have smaller footprint.

eSpeak is, indeed, just disturbing more than helping.

Would have to make new research on TTS engines at some time and see new developments in the area.

from pure-maps.

Thank you for the information.

• I did a little search on the web today and unfortunately there does not seem to be any new developments WRT Free Software TTS engines.
• PicoTTS is working fine (at least for me and except aforementioned issue): TTS quality and resource usage is good, plus it provides multiple languages. Although its quality is not really up to today's standards. Knowing it is in its "final" state, that's at least one thing less to think and care about. ;)
• While I considered trying Mimic again on an Xperia X (after failing to utilise it on two Jolla 1 phones), the facts that nobody described its output as exceptional and that it is "English only", make this "low priority (probably never get around to do that)" for me.
• What is an "AP model" in the context of speech synthesis?

from pure-maps.

AP model -> sorry, AP voice of mimic. Its actually quite good. As for developments, Mimic 2 may be of interest, but let's see if we get it ported.

from pure-maps.

Ah, missed that two days ago:
https://mycroft.ai/blog/mimic-2-is-live/
https://github.com/MycroftAI/mimic2
Thanks for the hint to look for "Mimic 2", that's really interesting to read.
Although it makes me a bit afraid of the resource use Mimic 2 may have.

from pure-maps.

Oh, I missed to mention the PicoTTS-fork nanoTTS for long.
Though nanoTTS focuses on improving pico2wav (i.e., its API, but not the TTS engine), it may be helpful.

from pure-maps.

Oh, you are already using nanoTTS (and may have implemented it).
Sorry for the unnecessary post (I should have looked more closely in the first place).

from pure-maps.

I wonder if we still have this instability and whether ability to test TTS engines already addressed main concerns with this.

from pure-maps.

I have not used this much lately, but it was working fine on my Jolla 1, when I did.
I also have not read any comments from others (e.g., at TMO) WRT this, anymore.

So, yes, closing seems to be the right thing.

from pure-maps.

Retested in real life situations last week. Unfortunately I have the impression that PureMaps' behaviour is still exactly the same as in @Pohli's description for PureMaps 1.7.0:

The only thing I've noticed in every version so far is that directly after starting a freshly calculated route there's no TTS for the "initial" directive, same after rerouting.

IMO it would be very helpful to emit via TTS "Voice navigation started" (or simply use the on-screen message text "Navigation on") rsp. "New route found" to ensure the user that no navigation instruction is missed.
Currently when the navigation mode with TTS is started, PureMaps stays silent, while there should be some audible feedback to make clear everything is working fine, because each time there are so many potential reasons for failing, e.g. low audio volume, failed BT connection etc. Better to provide users with the ability to realise that right at the start of an navigation, then later (while navigating).

P.S.: In the past I was able to hear a "New route found" sometimes after triggering it manually, but outputting this via TTS never was stable on my Jolla 1 phones. Now that always seem to fail, no matter how often I manually trigger a new route calculation while navigating with TTS on.

from pure-maps.

Good ideas! I'll look into it

from pure-maps.

Thanks a lot & will do, if necessary.

from pure-maps.