“When it comes to use in a screen reader, skipping words, or reading numbers incorrectly, is unacceptable.”

Modern AI TTS systems fail blind screen reader users who need fast, predictable voices at 800-900 words per minute. The problems: dependency bloat (100+ Python packages), accuracy issues like skipping words and misreading numbers, inability to stream audio in real-time, and lack of pitch/speed customization. Meanwhile, the industry-standard Eloquence voice from 2003 is becoming incompatible with 64-bit systems. A proper fix would require millions in funding and expertise spanning linguistics, signal processing, and audiology.