IBM Watson Text to Speech vs NVIDIA Magpie TTS

IBM Watson Text to Speech is an API cloud service that converts written text into audio in multiple languages and voices, for applications or watsonx Assistant. Deep neural networks trained on human speech produce its neural voices. SSML controls can change pronunciation, volume, pitch, speed, and other attributes; unusual pronunciations can be specified with IPA or IBM SPR. Listed speaking styles include GoodNews, Apology, and Uncertainty, and voice transformation can adjust strength, pitch, breathiness, rate, and timbre. The service supports nine languages and accepts up to 5,000 input units. Export formats include A-law, Basic audio, FLAC, Linear 16-bit PCM, MP3, Mu-law, Ogg, WebM, and WAV. Deployment options include public, private, hybrid, multicloud, and on-premises infrastructure. Deploy Anywhere is billed by contacting IBM. Premium can include a custom branded neural voice modeled on a chosen speaker using as little as one hour of recordings.

NVIDIA Magpie TTS has no write-up on Specifiction yet; its figures are above.