Save output to file

[pypingou]

Hi,

I would be very interesting if instead of playing the sound I could also use the library to save it to a file.
At the moment I have to use system call to do this and using pyttsx for this would be nice.

I wish I could provide a patch but this is beyond my python skills.

[Shenlok]

I'd also be interested in learning how to do this! I intend on using PyTTSx for text-to-speech on a webpage I am serving from a python script.

[kscottz]

I would love to have this feature.

[cereal]

Yeah me too - if you need any help just mail me. I can't find the right place to start this, so if you could tell me where the byte-or-whatever stream is 'created' I could do the rest ..

[parente]

I see two ways to approach this feature. One is to extend the API with a speakToFile method or some such exists and all the platform drivers implement it

A second would be to create a new driver per platform specifically for speaking to disk. This would double the number of drivers, but would keep the existing API.

In either case, code needs to be added per platform driver to output to disk. To the best of my knowledge, all three of the existing drivers (NSSpeechSynthesizer, espeak, and SAPI) have this capability. Finding and reading the doc is where it's at.

If someone has time to implement this feature and associated tests, I'll happily review a pull request. I don't have the cycles to work on it myself right now.

[pettarin]

Hi @parente .

First of all, thank you for writing pyttsx and its documentation.

I too would like a "write to file" API, instead of flushing the generated data at the output sound device, on Linux.

I have done some research on speak_lib.h, the public .h for the espeak lib, and I think there is no "pre-cooked" API for saving to file the produced wave data. It seems to me that one must write a callback function that flushes the data to file as it comes, with the proper WAVE header and ending flush. Essentially this means mimicking what the -w flag (= write to file) does in the src/espeak.cpp in the source of espeak.

(Full disclosure: I am looking for a lib-based approach to synthesize text in Python (or a C extension), to replace this code which uses subprocess to call the espeak executable: https://github.com/readbeyond/aeneas/blob/master/aeneas/espeakwrapper.py )