search

microbit-foundation / micropython-microbit-v2

Temporary home for MicroPython for micro:bit v2 as we stablise it before pushing upstream
MIT License
43 stars 24 forks source link

Speech quality #37

Closed dpgeorge closed 3 years ago

dpgeorge commented 3 years ago

The speech quality has recently been improved in 081bb4e7ecc418a768292367a8f8db3ffaf8d082 which adds double buffering to the output. Before this change the speech was very "rough" sounding, I think because there were gaps between the chunks going out to the audio mixer. With double buffering the speech output is much better and sounds the same as playing a whole speech sample as one big buffer.

Please test and compare with a micro:bit v1!

mytechnotalent commented 3 years ago

I will test @dpgeorge as I demo my Study Buddy featuring the bot called "Mr. George" named after you of course at the big micro:bit LIVE global conference on Friday - https://github.com/mytechnotalent/MicroPython-micro-bit_Study_Buddy here.

I will test this today. I will build from within the repo on this branch 081bb4e.

The issue I found was the speech was peaking or overdriving slightly.

mytechnotalent commented 3 years ago

@dpgeorge @microbit-carlos

Demo of change 1 - up close 

from microbit import display, Image
from speech import say
display.show(Image.SURPRISED)
say('Hello, my name is Mister George.')
say('It is so great to meet you all!')
say('I am ex si ted to help you change the world!')
display.show(Image.HAPPY)

FIXED_081bb4e.MOV.zip

SIGNIFICANT improvement! I included a video for you to review. Keep in mind it is MUCH better LIVE.

I included the compiled hex for you if you wanted to see for yourself. MICROBIT.hex.zip

microbit-carlos commented 3 years ago

Thanks Damien, I gave this a quick spin and sounds much better! And thanks Kevin for testing as well.

I've tagged and released beta.2 to deploy to the https://python.microbit.org/v/beta (still deploying) so it'll be easier for other people to test as well.