from Hacker News

Updates to Advanced Voice Mode for paid users

by mfiguiere on 6/7/25, 8:27 PM with 48 comments

  • by zaptrem on 6/7/25, 8:57 PM

    > Additionally, rare hallucinations in Voice Mode persist with this update, resulting in unintended sounds resembling ads, gibberish, or background music. We are actively investigating these issues and working toward a solution.

    Would be cool to hear some samples of this. I remember there was some hallucinated background music during the meditation demo in the original reveal livestream but haven't seen much beyond that. Artifact of training on podcasts to get natural intonation.

  • by kubb on 6/7/25, 8:56 PM

    I have the feeling that the Advanced Voice Mode is significantly worse than when I used it earlier this week. The voice sounds disinterested, and has weird intonation. It used to be excellent for foreign language conversation practice, now significantly worse.

    Edit: After using up my 15 minutes for testing, I have to say that the new voice is actually not bad, although I was used to something else. But it has a very clear "artificial" quality to it. It also sometimes misinterprets my input as something completely different than what I said, for example "please like my video and subscribe to my channel".

  • by TheTaytay on 6/7/25, 8:59 PM

    I wish they still had the voice mode that was _only_ text-to-speech, and speech-to-text. It didn't sound as good, but it was as smart as the underlying model. The advanced voice mode regularly goes off the rails for me, makes the same mistake repeatedly, and other things that the text-version of advanced LLMs hasn't done for months now.
  • by nickthegreek on 6/8/25, 1:41 AM

    They absolutely destroyed Sol. I’m not sure what it is now. the disinterest, the umms, the inability to speak directly to question, a new inflection but I am pretty mad. I am an avid voice user. I love to use the advanced voice while I’m doing tasks to explore new projects I want to work on and to get a basics understanding of home renovation tasks, etc. I had to finally change the voice to Maple but ran out of time to see if I could stand it. So disappointing.

    At least know I know i’m not crazy and there were in-fact changes rolled out.

  • by tallytarik on 6/7/25, 8:57 PM

    > Additionally, rare hallucinations in Voice Mode persist with this update, resulting in unintended sounds resembling ads, gibberish, or background music.

    This would be really funny if it weren’t real life.

  • by dedicate on 6/8/25, 12:23 AM

    In my daily use, I just want the answer, not a performance. I'd rather it sound like a smart assistant, not my best friend.
  • by mensetmanusman on 6/19/25, 1:02 PM

    Days late to the party because I thought it was based on poor internet. Arbor sounds totally stoned, lol, how did they release this?
  • by ed_mercer on 6/7/25, 10:17 PM

    I keep using standard voice mode (Cove) because I like its grounded voice a lot. The advanced Cove’s voice sounds too much like an overly happy guy. I wish I could tell it to chill and talk normally but it won’t.
  • by nikkwong on 6/8/25, 3:13 AM

    The women voices all sound like the valley girl that you wish wasn’t invited to the party. The male voices, sound well, similar to that I guess id say. I’d like voices that sound more like ethnic people found the crowds that many of us interlope in, rather than the pompous ivy-league educated girlfriend you wish your friend didn’t have. The product shouldn’t so clearly advertise that it was developed in a San Francisco monoculture.
  • by patwolf on 6/8/25, 1:02 AM

    I was using it earlier today and noticed something was different. It sounded more lethargic, and added a lot more "umms". It's not necessary bad, just something I need to get used to.

    I always get a laugh asking it to talk like an Ent, and I made sure to check that it could still do that.

  • by arnaudsm on 6/7/25, 10:44 PM

    If there's an OpenAI PM reading this: please add the model selector for voice modes. 80% of this thread is users confused about which model they're using.
  • by sandspar on 6/8/25, 4:10 AM

    I don't like how it laughs while it speaks. I associate this behavior with the anxious, neurotic middle class. It's discomfiting.
  • by cladopa on 6/7/25, 10:09 PM

    Today I used ChatGPT and the voice was disgusting for the first time since I use ChatGPT(months).

    It was the voice of someone(a woman) that was confrontational, someone who does not like you.

    It made me want to close and remove the chat immediately.