ElevenLabs offers advanced Text-to-Speech (TTS) capabilities with a variety of multilingual, natural-sounding voices. The following outlines key aspects of the ElevenLabs TTS service:
Language Support & Voices
ElevenLabs offers 20 default voices with various characteristics. You can explore and preview these voices from their documentation: ElevenLabs Default Voices or through their free web interface. The voices are multilingual, supporting 32 languages, and can automatically detect the language based on the AI agent's text.
Parloa leverages the real-time model by ElevenLabs, called flash_v2.5.
Voice Quality
Provides generative TTS with natural, expressive voices that are highly effective for personalized or high-quality interactions.
Latency
Slightly higher than Microsoft Azure, with response times ranging from 500-700 ms.
Availability & Pricing
ElevenLabs is generally available. However, it is currently an optional provider that has to be activated in your account.
Please contact Parloa Support or your Customer Success Manager to discuss using them in your project. ElevenLabs voices cost extra and need to be added to your contract and Data Processing Agreement.
Voice Control
Similar to OpenAI voices, ElevenLabs voices require custom text normalization for certain terms and numbers to ensure clarity in speech. SSML is not supported. Refer to OpenAI via Azure (Preview) for more details on common strategies to improve the results.
Processing Options
Processing currently happens in the US always. However, Parloa has entered into an enterprise agreement with ElevenLabs and assessed their data protection handling suitable for EU customers, too, making them a GDPR-compliant option nonetheless. The service does not retain any traces of inputs or outputs.
Integrating ElevenLabs
You need to agree to the processing of ElevenLabs for your project for us to activate it in your account. Reach out to your Parloa representative to kick-off the process.
Custom Voice
You can create your own voice with ElevenLabs as a Professional Voice Clone and integrate it into the Parloa project. Contact your Parloa representative for more details.
