Improve TCI session audio quality with Audacity

Hello, I apologize for my bad English.
I need advice and help to improve the audio quality of a recording of an instrumental transcommunication session
(TCI). I am a 54 year old woman who has no knowledge about sound techniques, I have obtained very interesting paraphonies, I have been able to identify some words, but other words are harder to understand because during communication the sound it usually goes down in intensity, and there are moments in which the voices are heard very faintly.

Is there, in this forum, a kind person who has used Audacity to enhance a recording obtained in an instrumental transcommunication session ?
I would really appreciate it if you could give me some advice.

First the good news: a free plugin for Audacity called LevelSpeech2 will make the quiet sounds louder.

Now the bad news: If by “paraphonies” you mean EVP, that’s just pareidolia.

The processing used in digital sound recordings makes them even more prone to make noise sound like speech than analog recordings, (e.g. tape).

I have done 3 sessions of TCI, 15’, 16’ and 12 `.
They have been long conversations, I asked security questions that were answered correctly. For example, I took a book in my hand, with a long title, and they began to answer it even before I finished the question. Also clear answers to personal and family questions.
At various times answers with long sentences consistent with the topic of conversation.
We have managed to understand, my daughter and I, most of what has been transmitted to us, But in some moments, there are words that the signal is weaker, less intense, it may be due to the variation in the intensity of the waves, for some reason that I do not know.
And there are few words that remain to be identified without error, the rest of the conversation is heard well and there are no doubts of interpretation.
The recordings have been listened to by several people, without being previously influenced, and all have agreed on the message that is heard, almost 90% of the message we have been able to interpret,we only need to interpret some words, we want to do it in the most objective way, aswe did with the rest of the sentences of the message.

The free plug-in for Audacity calledlevelspeak2 where do I find it? I thank you very much for your advice.

I did provide a hot-link to LevelSpeech2 in my post (above),
here it is explicitly … https://forum.audacityteam.org/t/improved-leveling-for-speech/33258/27

Audacity’s Native Compressor tool will also make quiet sections louder if make-up gain is engaged, (but IMO Levelspeech2 does a better job) …
Audacity's native compressor effect with make-up gain ticked.png
https://manual.audacityteam.org/man/compressor.html

There are plenty of examples showing EVP is audio pareidolia.
Digital audio processing, e.g. noise reduction, is particularly prone to accidentally creating voices which do not exist, e.g. https://freesound.org/people/Timbre/sounds/248088/

Here’s the science bit … https://youtu.be/MyQT78Bjt04?t=131

PS
I just remembered there is a plugin which is simpler to use than LevelSpeech2, which makes everything equally loud: called “Hyperexp” … https://wiki.audacityteam.org/wiki/Nyquist_Effect_Plug-ins#Hyperexp
It’s ideal if you want to listen to the noise floor, (but not much use for anything else).

Thank you very much for your answer.
I agree that erroneous interpretations must be avoided and an exhaustive and objective analysis must be carried out, to avoid fooling oneself with probable pareidolias of sound. That is, precisely, what I try to avoid.