Annoying side effect when changing the order of announcements
  •   There has been quit a debate about changing the order of announcements;
    distance first or maneuvre first.
      I am not well informed about all languages  but I think that western languages
    (I only tested English and Dutch) have the phenomenon that you can hear that
    the sentence comes to an end: the pitch lowers for the last syllable(s).  Then by
    only swapping the announcements you hear a sentence and there after an
    incomplete sentence!  Most voices have 2 versions for some announcements,
    a "sentence_ending" and a "non_sentence_ending" version.  So the last syllable(s0
    must also be reversed to have a natural sentence.
      To find out the right version of an announcement, it would be nice if you could
    "disassemble" the .MCA file to check the words (f.e. with VLC) of the different files.
    Then you can make a natural sounding sentence.
      If that could be done it also opens the possibility to construct your own
    "sounds_mine.mca", if you have your own sound files!
      I wonder of our "knowledgeable about" people can fix this for us.

    Regards knibja
  • 6 Comments sorted by
  • You mix up two things: recorded voices using media files, and TTS voices using texts to be spoken by the TTS engine.

    The latter will never work with VLC. The recorded voice are a "dead end".
    With TTS voices you will get (semi-natural) behvior nomatter the construction ofthe sentence because it will indeed lowerthe pitch at the end, no matter the order of the words. 

    (And don't use bold font face. You are "shouting")
  • It is however possible to change the order of the recorded messages in the same way as the TTS messages.
  • Yes of course. Thanks for the addition.

    What I meant though is that you can't make natural running sentences by recombining or even modifying the recorded voices.
    Whereas a TTS sentence will try to mimic the natural flow for a sentence in that language even though the sentence is grammatically nonsense.
  • Just to sum up and make it clearer:
    1) mca is zip-like file without public compiler/decompiler (I am not allowed to give you this tool)

    2) sounds_??.mca containts wav files

    3) tts.xml is replacement of wav files:
    e.g.
      <phrase>
        <token>A-KW-00-00</token>
        <text>keep left</text>
      </phrase>
    means replace A-KW-00-00.WAV with string "keep left"

    DO NOT ADD TOKENS IF THEY ARE NOT USED IN ORIGINAL XML
    (that would cause app crashing - it is not garanted that all waves are recorded in all languages)

    4) In navigation.xml is definition of order of messages:
    e.g.
        <manoeuvre>
          <!-- Roundabout. Take the [first, second, ... twentieth] exit -->
          <type>roundabout_now</type>
          <file>A-RO-00-00</file>
          <file>Cx-RO-00-01</file>
          <file>Cy-RO-,%EXIT_NUMBER,-11</file>
          <file>Cz-RO-00-10</file>
        </manoeuvre>

    I think I do not have to explain what it is doing as it is described in comment.

    tag manoeuvre defines messages for recorded voices and tts.

    If there is defined manoeuvre_tts - then it will be used for tts only
    (some messages have to be extra defined for TTS as there is not possible to annouce street name with recorded voice)
  • I know quite well the difference between recorded voices and tts. I thought that was clear
    because I mentioned a "sounds_mine.mca" with own sound files where tts uses text files.
    I use "nl_mobikom" as voice. When you simulate a route you can hear a difference
    between the pitches of the MANOEUVRE_ID at a certain distance and at the position
    of the manoeuvre. That is not by modifying a wav file but by using a different wav file.
    Simple as that.

    @hvdwolf
    I have no idea how my post is bold. It was certainly NOT my intention to use it.
    Being quite new here I have no idea how I could make the text bold. It is stupid that I not
    noticed that the text was bold but if I had I would not have known how to correct it.
    I apologize for the mistake.

    Regards knibja
  • I have a few comments about the vocal directions.

    When I depart from a calculated route, the app tries to return me to that route (which is what I expect), but sometimes the voice tells me to turn at an intersection that I have just passed. The reason for the delayed instruction appears to be that the voice is busy saying "Recalculating the route" when it should be saying "Turn Left/Right", etc. The turn statement occurs, but when roads are close together, that vocal direction is often given too late to make the turn.

    I don't think it should be necessary for the voice to say "Recalculating the route". It should be obvious the change has occurred from looking at the screen.

    When a route is first calculated, the voice says "Recalculating the route". It should say "Calculating the route" for initial calculation.

    It would be nice if the app would voice highway exit numbers. Here in the USA, the words "Take exit (number)" in (distance)" would be most useful, in my opinion.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In with OpenID

In this Discussion

Tagged