Change voice on Tablet for Speak to

[avpman]

I'm writing events to "speak" to the Android tablets I have throughout the house. How do I change the "voice" on the tablets? I found a TTS setting that lets me download, choose and set default voices, but that has no effect on what HS speaks to the tablet.

For the purposes of this exercise, I don't want to use any additional (i.e. Node-Red, Voice Monkey, Alexa. etc) software. I'll look into that later. Although, if there's an Andriod app that lets me change the voice on each tab, that would be cool.

Thanks in advance!

[Pete]

HS4 speaking to Android Tablets running Homeseer Touch play back a recorded TTS on the HS4 server whether running HS4 in Windows or Linux as far as I know. IE: TTS is sourced on the HS server running on Windows or Linux and one voice font for all HSTouch clients running on Android.

You can see the recorded wave files temporarily saved on the HS4 box. It is only the voice / speech configured on the HS4 server. IE: one for all clients.

In Windows HS4 it will be an MS SAPI voice and in Linux it is whatever TTS you use in Linux. Not sure today if Homeseer Touch uses native Android TTS.

Running HS4 on Ubuntu Linux here get one configured Linux TTS to the HSTouch application on Windows and native SAPI on speaker application at the same time.

Homeseer users please correct me if I am mistaken.

Here have never used any 3rd party speech proxies for Homeseer Speech and my tablets all run Windows HSTouch and speaker with built in SAPI speech. ALL of the 3rd party speech proxies use only the HS4 server speech source (recorded wave sounds).

Newer tinkering touchscreens here that run Windows 7 Ultimate / Windows 11 do run native Microsoft SAPI. (HP, Asus AIO, et al)

[avpman]

....You can see the recorded wave files temporarily saved on the HS4 box. It is only the voice / speech configured on the HS4 server. IE: one for all clients.
'
'
'

Ah ha, I found them! Yes, the crappy voice is in the wave files I found. Now the question is, how do I change the speech engine that HS uses in Linux? Havving it uses the same voice style for all tablets is no problem.

[Pete]

Most if not all of the Linux speech engines (native to the OS - not cloud based) use robotic sounding TTS.

There were efforts to implement SAPI speech on Linux way long time ago. Then with cloud based speech those efforts went away.

Many years ago I did get MS SAPI running on PlayonLinx.

Are you running HS4 in Linux Ubuntu with Mono on a Lenova Tiny?

[Pete]

I do it different today and utilize Windows Virtual boxes running on Ubuntu 22.04. That said all of my HSTouch clients run in Windows today.

Tested this a few years back and it worked. Today use PlayOnLinux on my Linux Lenova laptops to run MS Office and a few other Windows apps with direct links to my Ubuntu Desktop.

How to install and use Windows sapi5 voices on Linux

This would let you utilize SAPI5 voice fonts in native Ubuntu Linux.

[Guest]

The above is correct in that the HS voice is configured and generated on the system that the HS server is running on.

This is something I wrote up awhile ago to discuss how to set voices on HS3. Should still apply for HS4.

This is for a Homeseer server Windows 64 bit system running Windows 10. Other types of Homeseer server systems may behave differently. Voice processing is done on the server, not on the speaker clients (or tablets). So the voice configuration must be done on the Homeseer server system.

Homeseer HS3 is a 32 bit application, and only uses 32 bit voice files.

To see the 32 bit voice files, and to set the 32 bit default voice, use the file manager to double click on:

C:\Windows\SysWOW64\Speech\SpeechUX\sapi.cpl

Which will bring up an older style dialog box, with a drop down of the 32 bit voice selections available. Selecting a default voice here will cause Homeseer to begin using that voice as a default.

In the above dialect packages, it appears that only the US and UK version have 32 bit voices. David and Zira for the US, and Hazel for the UK.

There are other voice files available from third parties (usually for purchase), but be sure that they contain 32 bit voice files as well as the 64 bit ones.

This above sets up the default Homeseer voice. You can override these defaults by including XML (HTML like) tags in the speech text string, either in the Event Speak action setup, or in the hs.speak script call.

The XML tag for temporarily changing the voice is:

<voice required="name = Microsoft Zira Desktop">

The end tag is </voice>, but you do not need to include it unless you are getting fancy with your tags. The name to use is the name found in the 32 bit dialog box from above. You can actually include two voices in one voice statement, spell out words, and other stuff using XML tags.

The following is an example event action using Zira:

Then
Action: "Speak"
Text: <voice required="name = Microsoft Zira Desktop">This is an announcment

The following is an example of saying things in script, and using two separate voices:

  Dim buff As String
  buff = "<voice required=""name = Microsoft David Desktop"">Zira, I love you. I want to be with you forever.</voice>"
  buff = buff + "<voice required=""name = Microsoft Zira Desktop"">David, don't be an idiot. I can never love a mindless moron like you.</voice>"
  hs.Speak(buff, True)

[avpman]

Most if not all of the Linux speech engines (native to the OS - not cloud based) use robotic sounding TTS.

There were efforts to implement SAPI speech on Linux way long time ago. Then with cloud based speech those efforts went away.

Many years ago I did get MS SAPI running on PlayonLinx.

Are you running HS4 in Linux Ubuntu with Mono on a Lenova Tiny?

I upgraded the Lenovo tiny M93 to a M900 I7 with 16gb RAM. Debian 11.

[Pete]

Here too just purchased another m900.

I believe you still have a Russound zoned audio system.

While the embedded Windows tablets here run sapi I prefer to utilize the Russound for HS SAPI . sound.

Here have the audio output of the HS4 Ubuntu computer going to the Russound amp.

Mixing the audio output from one Alexa Dot there

Also running a tiny windows server on Ubuntu doing Homeseer speaker. Now 6 instances with a variety of sapi tts fonts.

[avpman]

Here too just purchased another m900.

I believe you still have a Russound zoned audio system.

While the embedded Windows tablets here run sapi I prefer to utilize the Russound for HS SAPI . sound.

Here have the audio output of the HS4 Ubuntu computer going to the Russound amp.

Mixing the audio output from one Alexa Dot there

Also running a tiny windows server on Ubuntu doing Homeseer speaker. Now 6 instances with a variety of sapi tts fonts.

I'll eventually set up something else, like V-Monkey or Node-Red to distribute the speech to Alexa devices. Right now I just want a quick and dirty way to announce some things on the tabs; with a little bit better voice.

"Here have the audio output of the HS4 Ubuntu computer going to the Russound amp."'
Are you using a better voice engine? I'm confused...

[zwolfpack]

If you're not completely cloud-adverse, you might consider the Amazon Polly voices.

I put this together based on the work of the thread's OP. Haven't used it in years, but just tried it and it still works.

https://forums.homeseer.com/forum/ho...91#post1250191

Idea behind it is to cache the phrases, so cloud calls only happen the first time a particular phrase is spoken.

[Pete]

Are you using a better voice engine? I'm confused...

The HS4 voice engine on the Ubuntu server is a Windows 32 bit 10Gb Virtual box running 6 instances of speaker.exe.

The Speaker instances connect to 2 Ubuntu HS4 boxes (one same box) and 2 Ubuntu HS3 boxes. I just point the speaker app to the HS4/3 servers running in Linux.

HS4 and Home Assistant and Oracle Virtual Box are running in Ubuntu 22.04 ===> linux audio out ==> Russound

===========> Same Ubuntu 22.04 ===> Oracle Windows Virtual box ===> SAPI Speaker audio ==> Russound
Amazon Dot Cloud audio out ==> is mixed with the HS4 audio out to Russound

Did a quickie drawing here using Visio (MS Office old) running on my Ubuntu Laptop via PlayOnLinux.



Doing same set up now for the HS4 box running on the Lenova M900 with 32 Gb of RAM and have moved the Windows VB to that box.

The multiple Amazon Alexa tabletop's and one dot are managed by Alexa Media Player running on Home Assistant which does way more than TTS.

Recently have added Google speech to Home Assistant.


This is the mixer I am using for the Dot audio and Ubuntu Audio.

Little Bear MC5 Mini Portable Stereo 4 Channel Audio Mixer Passive Live Studio Console Recording Low Noise (No Power Needed)

Very clean sound.