Spelling bee: 200 word-pronunciation clips on Sume TTS, one job each

Two hundred spelling-bee word clips of about 12 characters each cost $0.1140 on Sume TTS when you send one job per word, which is $0.000570 per clip at $0.0475 per 1,000 characters. One job with all 2,400 characters costs $0.1140 but returns one long file you would have to cut.

One job per word versus one long job

Price is linear in characters, so the cost is the same either way. The choice is about the output you want: separate files, or one file plus timings.

Per-word loop

Key each job by the word's list position and a version, so a rerun after a crash reuses finished jobs.

words = ["necessary", "rhythm", "conscientious"]
jobs = {}
for i, word in enumerate(words):
    jobs[word] = speak(f"The word is: {word}.", f"bee-2026-{i:03d}-v1")
for word, job in jobs.items():
    print(word, audio_url(job))

Gotchas

The carrier phrase counts toward usage; "The word is: " adds 13 characters per clip, so price the final string. For homophones, add a sentence of context so the intended word is spoken. Use pronunciation_dict_id for terms the voice gets wrong.