Announced on August 8, 2024, the speech synthesis models were trained on 45,000 hours of speech and let users control voice characteristics with text prompts.
On August 8, 2024, the release of Parler-TTS Mini and Large v1 was announced. These open-source text-to-speech models were trained on 45,000 hours of speech. According to the announcement, text prompts can control characteristics such as gender, background noise, speed, pitch, and reverberation.
The announcement says weights, training data, and recipes are available for engineers to evaluate controllable TTS models. To verify the details, consult the original announcement and the materials provided, checking their scope and terms of use before testing. If you use AI to study or apply the material, avoid sending personal data or confidential content until you have checked the chosen tool’s policies and safeguards.