Why it matters for your business
In a voice agent, TTS is what callers actually hear, so voice quality, pronunciation and pacing shape their impression of your business. Streaming TTS keeps conversations responsive by starting playback before the full reply is written. Pronunciation of names, addresses, product codes and numbers often needs tuning. Many providers offer voice cloning, which raises consent and impersonation concerns; use only voices you have clear rights to, and be open with callers that they are speaking with an automated system.
Example
A property management company's after-hours maintenance line uses TTS to speak the voice agent's replies. During testing, the voice mispronounced several apartment community names and read unit numbers like 2B as words. The team adds a pronunciation dictionary and formats unit numbers for speech before sending text to the TTS engine.