Added an option of output: the wave array directly to memory (as a list of number 16-bit) #96
Conversation
…_to_file in linux
The pypi link was to pypi, not to the pyttsx project page on pypi.
- the platform specific dependencies are already handled by `extras_require` - when the wheel is built on macos / windows the `install_requires` will contain extra values (it appears the one on pypi for the latest version was built on macos)
add to_memory function
WARNING, only implemented for windows
usage:
>>> engine.to_memory(ss, audio_array)
>>> engine.runAndWait()
>>> data = np.expand_dims(np.asarray(audio_array[0] ,dtype=np.int16),axis=1)
>>> # you can test the output by: scipy.io.wavfile.write('name.wav',22050,data)
redreamality
changed the title
redreamality
changed the title
|
@redreamality This will most likely not be merged until the conflicts are resolved in your repository. |
|
@redreamality @nateshmbhat i commited it here: https://github.com/Jiangshan00001/pyttsx4 |
|
@Jiangshan00001 can u raise a PR here for the same? |
|
sure @nateshmbhat |
|
Hi @Jiangshan00001 . there are conflicts in this PR. would you be open to resolve the conflicts and update the PR. would be glad to merge it. |
|
conflicts resolved |
|
I see now this and PR #270. Pretty similar @redreamality its been a while. Whats your thoughts on save_to_memory. This doesnt sit with me right as a method name.. its not really in memory - its a save_to_bytes isnt it? Hold on - I've looked again and no , i see what you are doing. Thats pretty specific. What about if this module did have a save_to_bytes. Then in your local script you would be able to just pile up the PCM 16 bytestream however you want. Wouldnt that be more useful? NB: It would be a bit like this: https://github.com/willwade/tts-wrapper?tab=readme-ov-file#2-real-time-playback-using-sounddevice |
|
FYI.. Ive been digging into this more. SAPI does have a SAPI.SpMemoryStream method as @Jiangshan00001 has done.. But in Mac I cant find a native way of doing this so you would have to write a hack of writing to a temp file and then parsing it out. or using more frameworks like AVFoundation.. eSpeak should be easy to do.. I would push my PR but its My branch is here https://github.com/willwade/pyttsx3/tree/to_bytestream If I figure out even a hack method for nsss I'll do a PR |
I'm just providing a method that keeps the stream in memory, without the need of any temp file> FYI.. Ive been digging into this more. SAPI does have a SAPI.SpMemoryStream method as @Jiangshan00001 has done.. But in Mac I cant find a native way of doing this so you would have to write a hack of writing to a temp file and then parsing it out. or using more frameworks like AVFoundation..
I'm just providing a method that keeps the stream in memory, without the need of any temp file. This method, as i mentioned at very beginning, just works for windows. |
|
Yeah. I've done it now for espeak too now. I think if we can do a method for nsss it's worth reviewing this. I'll push it as a WIP for review |
CAUTION. only implemented for Windows sapi5 driver.