Right. These machine voices are difficult to hear in the first place. They should use a man's voice for masculine and female voice for feminine so that we don't get a wrong answer from mistakenly assuming she said cansada.. Or at least count it as a typo rather than making it wrong
No, you were right. Sueño is the noun for sleep (although it is also the noun for dream and the conjugation of verb soñar corresponding to 'I dream').
Duerme is not a noun, it is the conjugation of verb dormir corresponding either to 'he/she/it/[formal you] sleep(s)' or to 'sleep! [informal you]' as an imperative form.
English requires the infinitive (to....) after "want', if a verb is going to follow the "want."
I want to leave; I want to win the game; I want to sleep. (Here, "sleep" or "to sleep" is a verb.
Quiero irme; quiero ganar el juego. Quiero dormir.
I want for nothing (no verb follows). I want my book. I want everything, said Trump.
No quiero nada (no sigue ningún verbo). Quiero mi libro. Quiero todo, dijo Trump.
In my sleep (noun), I dream. En mi sueño (sustantivo), sueño.
When you lie down, you will not be afraid. Your sleep (noun) will be sweet.” – Proverbs 3:24. (Cuando te acuestes, no tendrás miedo. Tu sueño será dulce")
“A well spent day brings happy sleep (noun).” – ("Un día bien empleado trae un sueño feliz.") -- Leonardo de Vinci.
Sueño (verbo) con el día en que pueda dormir (infinitivo, verbo). (I dream of the day when I can sleep.)
Remember: All verbs start as an infinitive (i.e. hablar, comer, dormir) meaning "to speak", "to eat", and "to sleep", respectively. The conjugations come from those, with the personal pronoun. I.e. "Yo hablo" i speak, "yo como" I eat, and "yo duermo" I sleep. (Dormir is an irregular verb - but you'll get to that later.)
Use either one, it's not a matter of grammar. The verb ending already identifies the subject, so the pronoun can be omitted.
We native Spanish speakers tend to drop 'yo' most of the time and use it mostly for emphasis. As long as you don't say it twice in a sentence it won't sound too weird, though.
A female voice - should NOT be reading male parts (& vice versa). She should automatically switch to her natural case. It's hard enough to understand final vowels in rapid speech clearly, without such bewilderingly misleading twists! These things lead to undue stress, and slower progress, without any corresponding benefit.