Build a Songwriting Assistant with Automatic Syllable & Stress Pattern Validation

Please read the entire brief carefully before applying. I am a songwriter and I write songs in English with the help of ChatGPT. I am looking for a developer to build a simple tool that improves the lyric-writing workflow by automatically validating syllable counts and stress patterns before showing lyric suggestions to the user. The problem I am trying to solve is very specific. When writing lyrics, I often work with a predefined melody. For a given section of a song, I already know: what the lyrics need to say (meaning/content); the rhyme scheme; the required syllable count for each line; the required stress pattern for each line. For example, I might ask: "Write 4 lines for the first verse. The verse should describe a little boy sleeping alone in his bedroom at night. Meaning/content: line 1: introduce the boy alone in his room line 2: describe the quiet bedroom line 3: suggest innocence or vulnerability line 4: end with a strong visual image Formal constraints: rhyme scheme: ABBA line 1: 8 syllables, specific stress pattern line 2: 10 syllables, specific stress pattern line 3: 10 syllables, specific stress pattern line 4: 8 syllables, specific stress pattern" Current LLMs usually understand the meaning, imagery and rhyme requirements reasonably well. The main problem is that they very often produce lines with the wrong syllable count and/or the wrong stress pattern. In practice, this makes the results unusable because the lyrics must fit an existing melody. To be usable, each line must match the syllable count and stress pattern defined by me. What I would like is: a conversational songwriting assistant; one conversation per song project; memory of the song being worked on; the ability to ask for lyric suggestions at any point in the conversation; automatic validation against a formal template defined by me; for each line, I define the required syllable count and stress pattern; any line that does not match those requirements must be rejected automatically; only valid results should be displayed. I have already found an existing open-source phonetic engine called Phonetik / phonetik-js, which appears capable of: counting syllables; identifying stressed and unstressed syllables; analyzing the complete stress pattern of a line; comparing sounds and rhymes; providing phonetic information for words. I do not know whether this engine is sufficient for the task, but it may significantly reduce development time. Important: I am not looking for a generic AI chatbot. I am specifically looking for a solution that can reliably validate whether generated lyrics match a user-defined syllable count and stress pattern. If you apply, please explain: how you would approach the validation problem; whether you would use phonetik-js or another phonetic engine; how you would test accuracy; what level of reliability you believe is realistically achievable; your estimated cost and timeline. To confirm that you have read the entire brief, please start your proposal with the word: BEACH I will automatically ignore proposals that do not start with this word.