Announcement

Collapse
No announcement yet.

Caching Speech routines using Amazon Polly

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • zwolfpack
    replied
    Originally posted by koburg View Post
    so it is not possible to use polly on windows with homeseer at the moment ?
    No, on Windows the speech setup is totally different. They don't provide a "hook" like the speak_to_file.sh script.

    I bought a package of Ivona voices a few years back, from here (iirc). I think Amazon may have absorbed Ivona, and now those voices are available as Polly voices.

    Leave a comment:


  • koburg
    replied
    ok thanks ill try

    so it is not possible to use polly on windows with homeseer at the moment ?

    Leave a comment:


  • zwolfpack
    replied
    I have a test installation on a raspberry pi, that I used to help Timon develop part of the PollyC package. My real HS is running on Windows, which doesn't run PollyC of course.

    See https://linuxize.com/post/how-to-ins...-ubuntu-20.04/

    Try editing the Makefile and on line 50, delete "python-pip", leaving only "python3-pip". Python2 apparently isn't supported anymore, but PollyC will use python3 instead.

    You make need to delete line 54 as well, not sure.

    Leave a comment:


  • koburg
    replied
    zwolfpack are you running homeseer on ubuntu ?
    if desktop or server ?

    my setup with the working polly is RPI 3 and the homeseer image for raspberry (5 plugin)

    i have tried setup a VM with ubuntu 20.xx desktop

    but when I want to install polly I get this

    Code:
    E: Unable to locate package python-pip make: *** [Makefile:50: install-pip] Error 100
    i have installed and update both python 3 and 2 and also done maybe 10 guides on how to fix this "PIP" error.

    but still same errors.

    Leave a comment:


  • koburg
    replied
    in homeseer setup : DK Europe/copenhagen in Polly setup eu-central-1

    but again not sure if it was the solution maybe I try change back later. to check.

    Leave a comment:


  • zwolfpack
    replied
    Looks good.

    Interesting about the region. What were you using before that didn't work?

    Leave a comment:


  • koburg
    replied
    all right now it is working before you answer I also tried change my location us/newyork in homeseer

    not sure if it did the trick and change another region in the polly config. (yours)

    and I saw the error about "=" but it did not work before where I had the right format.

    but do now.

    Code:
    Region: us-west-1
    
    Key ID: xxxx
    
    Key: xxxx
    
    Text: rødgrød med fløde
    
    Voice: Naja
    
    Format: mp3
    
    Text Type: text
    
    Polly Output: /tmp/speak_to_file1491.mp3
    
    Cache File: /usr/local/HomeSeer/./PollyCache/6e6d0612e4f6adc98ee3a072426fb52ddcdb0673086cb3212f65428a2e46 ca1a69c2dfcb9d95e7143f57ccd8092fc19dcbe0673f63c7bd41703b90db 267bdef7.mp3
    
    Cache Hit: False
    
    /usr/local/HomeSeer/html/Sonos/Announcements/Ann_ALL_0.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, stereo 44100 Hz

    Leave a comment:


  • zwolfpack
    replied
    This indicates it's having trouble encoding the text string (which it does to create the cache hash key). What phrase are you trying to send?

    Also, appears that the initial voice-id tag is malformed:

    Code:
    Warning: metatag <voice-id"mads"> ignored
    Format for this should be
    Code:
    <voice-id="mads">
    looks like it might be miising the "=" character.

    Leave a comment:


  • koburg
    replied
    alright got it do not seem to be good

    Code:
    pi@HS3Pi3v6:/tmp $ cat speak_to_file.log
    
    Warning: metatag <voice-id"mads"> ignored
    
    Traceback (most recent call last):
    
    File "/usr/local/HomeSeer/./PollyC.py3", line 194, in <module>
    
    main(sys.argv[1:])
    
    File "/usr/local/HomeSeer/./PollyC.py3", line 142, in main
    
    hashobj.update(hashtext.encode('utf-8'))
    
    UnicodeEncodeError: 'utf-8' codec can't encode character '\udcc3' in position 18: surrogates not allowed
    
    /usr/local/HomeSeer/html/Sonos/Announcements/Ann_ALL_0.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 16000 Hz

    Leave a comment:


  • zwolfpack
    replied
    Referring back to post #75, first step is to execute

    Code:
    touch /tmp/tts.debug
    This command creates an empty file /tmp/tts.debug. From your posting of the contents of /tmp, this file is absent. Run that command, trigger the speak action, then check /tmp again and you should see the file 'speak_to_file.log'. Post the contents of that, except for the key info.

    Note that the /tmp directory gets cleared by the Linux operating system upon reboot, so anything there will be gone once you do that.

    From the other commands, the installation looks good.

    Leave a comment:


  • koburg
    replied
    this is what I got from the commands

    Code:
    pi@HS3Pi3v6:~ $ ls -l /tmp
    
    total 120
    
    -rw-r--r-- 1 root root 118784 Jul 5 17:06 HomeSeerLog.hsd
    
    drwx------ 3 root root 4096 Jul 5 13:17 systemd-private-b7dcb8e9e63d42af806c57becbc1286a-systemd-timesyncd.service-Fucqfd
    
    pi@HS3Pi3v6:~ $ cd /usr/local/HomeSeer
    
    pi@HS3Pi3v6:/usr/local/HomeSeer $ ls -ld PollyC*
    
    drwxr-xr-x 2 root root 4096 Jun 30 14:49 PollyCache
    
    -rw------- 1 root root 205 Jun 28 12:50 PollyC.keys
    
    -rwxr-xr-x 1 root root 7886 Jun 28 12:06 PollyC.py
    
    -rwxr-xr-x 1 root root 7883 Jun 28 12:06 PollyC.py3
    
    pi@HS3Pi3v6:/usr/local/HomeSeer $ ls -l speak_to_file.sh
    
    -rwxr-xr-x 1 root root 1840 Jun 28 12:06 speak_to_file.sh

    Leave a comment:


  • zwolfpack
    replied
    The files under user pi aren't used after they're installed into the HomeSeer folder.

    Run these commands and post results

    Code:
    ls -l /tmp
    
    cd /usr/local/HomeSeer    # or where you've installed HomeSeer
    ls -ld PollyC*
    ls -l speak_to_file.sh

    Leave a comment:


  • koburg
    replied
    I tried run the command

    Code:
    touch /tmp/tts.debug
    I can not find any files.

    and when I log in to the RPI as pi@192.xxxxxxxxx I only see PollyC folder.

    and this is the files in folder

    Code:
    01settings LICENSE Makefile PollyC.py PollyC.py3 README.md speak.sh speak_to_file.sh

    Leave a comment:


  • zwolfpack
    replied
    My region is 'us-west-1'.

    To enable debug, execute this Linux command
    Code:
    touch /tmp/tts.debug
    Then run the speak event. Look for the file /tmp/speak_to_file.log and see if there's any indication of the problem. (Make sure to delete key info prior to posting).

    Here's mine:
    Code:
    Region: us-west-1
    Key ID: ---
    Key: ---
    Text: øæå
    Voice: Naja
    Format: mp3
    Text Type: text
    Polly Output: /tmp/speak_to_file10118.mp3
    Cache File: /opt/HomeSeer/./PollyCache/2226a16e82efdec6b79c17d2300c482e22a534887944363940a1e56915377a456e34d3782f984bd0d191f0838d86075389c4fdff8cbaa1f2de99150fae98da06.mp3
    Cache Hit: False
    /opt/HomeSeer/html/Sonos/Announcements/Ann_LVR_0.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, stereo 44100 Hz
    Attached Files

    Leave a comment:


  • koburg
    replied
    Hmm Strange because when I use øæå in a phrase it Will not Play correct.

    What about setup what Area code are you using. With danish voices ?


    Sent from my iPhone using Tapatalk Pro

    Leave a comment:

Working...
X