Acute accent lost from speech output

[Veli Laasala]

I'm using a SAPI 5 TTS. When text input includes an acute accent
character (´), my TTS receives a space character instead. Symbol
preferences seem to have no effect on this. When I replace this
character in a user dictionary with for example, apostrophe, the
replacing character is received by TTS. So, the acute accent is lost
after user dictionaries, but where?

NVDA versions 2025.3.3 and 2026.1 beta.

Thanks in advance,

Veli

[Mohamed Al-Hajamy]

Go to NVDA's Punctuation/symbol pronunciation dialog, and in the filter
by field, type the acute there and it should show up.

From there you can choose whether the character is sent to the
synthesizer with the send actual symbol to synthesizer option.

[Veli Laasala]

As I said in my initial post, preferences in the Symbols dialog of NVDA
has no effect on this. I also said that the acute accent character is
send to user dictionaries, because I can replace it there with another
symbol. I want the original symbol (´) to my TTS, but get it replaced
with a space character, for example "it´s" becomes "it s". Where does
this replacement happend after user dictionaries have been processed,
and why acute is replaced with a space?

Thanks in advance,
Veli.

[Veli Laasala]

It seems the Unicode normalization does this unwanted conversion from
"´" to " ".

Regards,
Veli

[Cyrille BOUGOT]

Hello

In NVDA, text to be spoken is processed as follows: dictionarie proicessing, then punctuation and symbols processing and then unicode normalization.

So if you modify the character in one of the dictionaries, the original symbol won't just go further.

I have tested modifying the "Punctuation/symbol pronunciation" and can check that the character "´" is transmitted to the synth, at least at the level of the speech manager log, which happens after dictionaries, symbol/punctuation processing and normalization.

Can you double check in the symbols/punctuation dialog?

I have level set on "most", but you can do as you want, and overall "Send actual symbol to synthesizer" set to "only below symbol's level" instead of the default "never".

What are your settings for this character?

Cheers,

Cyrille

[Veli Laasala]

Cyrille BOUGOT kirjoitti 27.4.2026 klo 14:06:
> In NVDA, text to be spoken is processed as follows: dictionarie
> proicessing, then punctuation and symbols processing and then unicode
> normalization.

Ok, this is the order I've expected.

> So if you modify the character in one of the dictionaries, the original
> symbol won't just go further.

This is absolutely clear for me.

> I have tested modifying the "Punctuation/symbol pronunciation" and can
> check that the character "´" is transmitted to the synth, at least at
> the level of the speech manager log, which happens after dictionaries,
> symbol/punctuation processing and normalization.

Then, how can you explain, why acute accent is received by my synth only
when the Unicode normalization is turned off?

> Can you double check in the symbols/punctuation dialog?
> I have level set on "most", but you can do as you want, and overall
> "Send actual symbol to synthesizer" set to "only below symbol's level"
> instead of the default "never".
> What are your settings for this character?

My settings for this character are as follows: level "most", and send to
synth "only when below symbol's level". And my overall symbol level is
"some".

Another NVDA user with similar settings has confirmed, that
normalization replaces acute accent with a space.

Veli

[Cyrille BOUGOT]

Hi Veli

My bad, seems that I had not tested correctly the first time.

I have re-tested and am able to reproduce your issue too.

Can you open a GitHub issue for that on NVDA's GitHub repository?

Thanks.

Cheers,

Cyrille

[Cyrille BOUGOT]

Alternatively, I can open the issue for you. I'll think I'll do it by the wy, while I still have the issue clearly in mind and my first findings.

But can you explain, in the first place the impact for you? How does your SAPI TTS work with acute accents and why is it so important for it / you? This will allow NV Access to give a priority to the issue. Thanks.