Can narrator.device output to raw file?

[Nightfox]

Hi

I was wondering for a small project whether it would be possible to get the amiga’s text to speech (narrator.device) to output to a raw audio file instead of being played through the speakers or is this just an impossible task?

I was wanting to create a program to do this so these files can be exported later to PC to be used in a modern app. Ideally have amiga emulator running in the background with some automation to get the files over to the PC

Final goal: on windows i can type something and the text gets processed on amiga and then the PC receives the speech file and can play it back

[Zack]

Could this be of any use? http://aminet.net/package/util/conv/say2raw
Have not tried it myself.

[Nightfox]

Originally Posted by Zack

Could this be of any use? http://aminet.net/package/util/conv/say2raw
Have not tried it myself.

Thanks. I'll try this one in the meantime but looking at the doc, it has a fixed buffer of 100000 bytes so if a speech is too long for that, it results in an error so it's limited.

[Thomas Richter]

Originally Posted by Nightfox

I was wondering for a small project whether it would be possible to get the amiga’s text to speech (narrator.device) to output to a raw audio file instead of being played through the speakers or is this just an impossible task?

The narrator.device output goes to the audio.device, always. This is hardcoded. What one could do is patch up the audio.device BeginIO() function and by that redirect the output to a file, but there is no option or flag of the narrator.device itself.

[Nightfox]

Originally Posted by Thomas Richter

The narrator.device output goes to the audio.device, always. This is hardcoded. What one could do is patch up the audio.device BeginIO() function and by that redirect the output to a file, but there is no option or flag of the narrator.device itself.

Interesting. I'm assuming there's no modern port of narrator.device that works on x86 is there? That would be the most convenient option.


Otherwise, is there a commented disassembly of audio.device as a starting point for patching that function?

[hooverphonique]

If you are doing this purely to get the "sound" of the amiga narrator, maybe you could find an x86 implementation that generates the same sound, considering C= licensed the speech tech from a third party (can't remember who).

[Nightfox]

All I need is the sound. I have an existing text to speech iOS app that I made and I just thought it would be cool if possible to get the old Amiga voice in there as a selectable voice. Even if it means sending the text to a server to process the text which sends back the audio file to be played back by the phone.

[Thomas Richter]

Originally Posted by Nightfox

Interesting. I'm assuming there's no modern port of narrator.device that works on x86 is there?

There certainly is:


http://www.text2speech.com/


Though this is commercial. Note that this product is by the same people that created the narrator.device. I do not know how well this works, and I am not affiliated with them.

Originally Posted by Nightfox

Otherwise, is there a commented disassembly of audio.device as a starting point for patching that function?

Why would you need a dissambly for that? The device interface is publically available.