Speech slows down when using espeak

[NSoiffer]

The speech gets slower and slower when espeak is used and the MathCAT rate is set to be slower than regular speech. There is no problem with other speech synthesizers.

Details of the bug can be found in nvaccess/nvda#15221.

[bhavyashah]

This is exactly the bug I was referring to. Is your understanding that this bug has been patched and is no longer experienced by a user? If so, I think I still face it.

[NSoiffer]

I thought it was fixed. It didn't happen to me in all the testing I did for the other bug. Is the other bug fixed? Do you have a reproducible example for this bug?

…

On Tue, Oct 31, 2023, 8:50 AM bhavyashah ***@***.***> wrote: This is exactly the bug I was referring to. Is your understanding that this bug has been patched and is no longer experienced by a user? If so, I think I still face it. — Reply to this email directly, view it on GitHub <#51 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALZM3FKQOWGFGQV37JQYXDYCEM4ZAVCNFSM6AAAAAA25NM4QKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBXGQ4TKNBTHE> . You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[bhavyashah]

In my limited testing with the new build, cutoffs are fixed and slowdowns are not (they may have worsened). ESpeak-NG at 35% with rate boost enabled and relative speech rate at 60. Variant Steph and language set to British English and NVDA version 2023.3 if that matters. Some examples below:
$X_1, X_2, \ldots, X_n$
$Y_n=\max \left{X_1, \ldots, X_n\right}$
$E\left(Y_{\mathrm{n}}\right)$
$P(0.1&lt;\bar{X}&lt;0.2)$
By "worsened", I feel like the slowing down begins sooner in the Math expression than in the previous build. I may be imagining this difference too so don't quote me on this.

[NV-Codes]

Although I do not use it as my primary synthesizer, I was experimenting with eSpeak and noticed this issue (MathCAT 0.6.8 RC 2). It seems that eSpeak progressively slows down while reading math, and the speech for surrounding text is affected as well until one navigates to a paragraph of text only (having no math). It then returns to its typical speed, but its behavior is not like the OneCore voices'.

[NSoiffer]

Reopening. I believe there was a change in NVDA to espeak voices recently, although that may be with 2025 alpha.

[NSoiffer]

I tracked the problem down to the pitch change caused by the capital letters. This is not a MathCAT bug, but an espeak bug. I "monkey-patched" espeak earlier, but I've modified the patch with a work around. More details are in this NVDA bug report.