Systems and methods for generating multi-language media content with automatic selection of matching voices

A computer-implemented system for automated voice casting comprising:

• a computer processor;
• a voice sample processing neural network in communication with the computer processor;
• and
• a computer memory having instructions stored therein that, when executed by the processor, cause the processor to:
• retrieve from a media library a primary audio sample comprising speech of a primary speaker in a primary language;
• generate a primary voice sample for the primary audio sample, the primary voice sample including utterances meeting at least one predetermined primary utterance type criterion;
• determine using the neural network an embedding for the primary voice sample, the embedding providing a multidimensional numeric representation of acoustic features of the primary voice sample;
• retrieve from the media library a plurality of candidate audio samples each comprising speech from a respective candidate speaker in a target language;
• generate a respective candidate voice sample for each candidate audio sample, each of the candidate voice samples including utterances meeting at least one predetermined candidate utterance type criterion;
• determine using the neural network a respective embedding for each candidate voice sample, each respective embedding providing a multidimensional numeric representation of acoustic features of such candidate voice sample;
• generate for each candidate voice sample a similarity score relative to the primary voice sample, the similarity score being a function of the embedding of such candidate voice sample and the embedding of the primary voice sample;
• and
• identify a specific candidate speaker as providing a vocal match to the primary speaker based on the similarity scores.