Announce Each Track Before It Plays?

[ebs]

I would like to use a text-to-speech (TTS) synthesizer program that I wrote to announce each track before it is played. My basic thoughts are:

1. Use "Player.onTrackChange" to determine when a new track will be played.
2. Use "Player.pause" to pause.
3. Use the "Player.getArtistAtPos", "Player.getAlbumAtPos" and "Player.getTitleAtPos" to get the track information.
4. Use "Program.run" to run my TTS program to speak the track information.
5. Use "Player.play" to start the track playing.

Can anyone point out any problems with this approach?

Or can the Silverjuke team make my life MUCH easier by putting this function directly into the program? What can I say - I'm lazy!

Thanks for any help!

Regards,
Eric

[Service-Team]

ebs wrote:
> Can anyone point out any problems with this
> approach?

In common, your approach should work.

However, instead of I would prefer and check once a second if a playing track is "close to end". If so, I would create and enqueue the TTS file for the next track (and remember that I've created a TTS file for the given queue position).

> Or can the Silverjuke team make my life MUCH
> easier by putting this function directly into the
> program? What can I say - I'm lazy!

Well, we do not have a TTS system, so I do not think a TTS feature will go to the "core" the next time ...

Best regards,
Your Silverjuke Team

[ebs]

Thanks for your fast response! I will consider your suggestions carefully. As far as a TTS system, there's SAPI built into Windows - that's all I'm using. Will you consider putting it into the core system now?

[Service-Team]

ebs wrote:
> Will you consider putting it into
> the core system now?

Not really We think, this is a rather special feature and a script or a module are a good place for this. Moreover, we'd like to leave some fun for script testing.

BTW: Is the SAPI for english fine? The german pronunciation was not very good we tested this the last time.

[ebs]

Quote:

> BTW: Is the SAPI for english fine? The german pronunciation
> was not very good we tested this the last time.

It depends very much on the particular TTS engine you use. I have found the AT&T Natural Voices to be pretty good in English and French, but I don't know about German. You can try it out at http://www.research.att.com/~ttsweb/tts/demo.php using the Klara or Reiner voices.

For English, I think that the NeoSpeech voices are absolutely the best. I created a telephone response system for a client with these voices and they tell me that people always try to have a conversation with the TTS voice - it's that convincing!

Regards,
Eric

[Service-Team]

Thank you for the demo link. Some english text like

This was "The Beatles: Yellow Submarine". Next track is "Rolling stones: Ruby tuesday".

sounds rather good. Some text in German ...

Dies war "Joachim Witt" mit "Der goldene Reiter". Der nächste Titel ist: "Rolling stones" mit "Ruby tuesday".

not too bad. Of course a big problem for the text above is the mixture of languages. Maybe one can simply use the english text or I think it should also be possible to switch the language in a text.

Another problem could be to get the "raw" wave data after it is created but before it gets to any speaker. This would be needed for a more exact processing and for more complicated output systems. Do you know if this is possible with SAPI?

However, all in all, this is an interesting project.

[ebs]

> Another problem could be to get the "raw" wave data after it is created but
> before it gets to any speaker. This would be needed for a more exact
> processing and for more complicated output systems. Do you know if this is
> possible with SAPI?


Yes - there are some examples here:

http://msdn2.microsoft.com/en-us/library/ms717065.aspx

Regards,
Eric

[Service-Team]

Thanks again.

Best regards,
Your Silverjuke Team

[ebs]

Another question, please:

When Silverjuke is paused, can another program use the sound hardware? In other words, can I have my TTS application directly speak the track information, instead of creating a .wav file and queuing that in Silverjuke? Is this the same for everyone, or is this something specific to my computer or Windows setup?

Is there an advantage to doing it one way or the other? In either case, I would check periodically until within a few second of the end of a track, then run my program.

Regards,
Eric

[Service-Team]

Hello Eric,

if you "Pause" Silverjuke, the hardware is not released, instead only "0" values are written to the output which results in silence then.

However, if you use at least Windows 2000, sharing the sound hardware by different applications should be no problem. Normally, it is even possible to run different audio outputs at the same time.
Depending on the used souncard/driver, it may be necessary to disable the "Use hardware" option at "Settings / Playback / Devices".

> Is there an advantage to doing it one way or the
> other? In either case, I would check periodically
> until within a few second of the end of a track,
> then run my program.

One advantage if Silverjuke plays the TTS file is that this will go though Silverjuke's volume and FX controls. If this is not needed, I do not see an advantage for the one or for the other method.

Best regards,
Your Silverjuke Team