I use Ember and the new language model not only sounds very bad, but also thinks very superficially and childishly.
A big problem is that it often emits the information “I can’t talk about it” even though we talked about the weather.
The new model… as if it does not read the memory and does not analyze the knowledge about the client. It treats me as if it just met me. Of course, when I remind him to analyze his memory, he does so, but he still continues the conversation like a salesperson in a store.
Same with me. Have some concerns about the newer voice model. Compared to the older version, it feels less natural and engaging. The slower pacing of the older model allowed me to process my thoughts, while the newer one interrupts or rushes the conversation, making it feel less thoughtful.
The lack of natural pauses and “umms” in the newer model reduces the sense of being in a real dialogue. These elements made the older model feel attentive and lifelike, as if it was “thinking.”
I also preferred the deeper tone of the older voice, which felt more soothing and emotionally engaged. It often encouraged me with phrases like, “Great question, Agata,” or “That’s such a deep thought.” This added warmth and made the interaction feel supportive.
In contrast, the newer model feels transactional, like a service person in a hurry. I hope future updates bring back the patience and connection I valued in the older model.
1 Like