Perhaps, KaniWani with some modifications? (Like editing this UserScript – KaniWani Audio - #10 by rymek)
Otherwise, some websites with audio segment shadowing.
It’s better to hear in context, anyway. So,
Vocabularies are in /resources/**/data.ndjson of "word_base_list" key, and sounds are in "sound" (URL can be figured out from the UserScript in the first post of the topic). So perhaps, ensure that you know all of the words in the "sentence" first.
Otherwise, getting other series’ subtitle (transcript) files and their audio, should be possible. YouTube (some of them has native JP sub) is also technically possible.