Yes, let's keep our implosion drama separate from our product announcements shall we?
Changed from https://twitter.com/gdb/status/1727067288740970877 above, plus I degregged the title. I'm sure he won't mind. (submitted title was "ChatGPT Voice for All Free Users Announced (By Greg Brockman)")
It seems much better now too. Before, it would do exactly one query and read one or two results and then give up. Yesterday I asked a question and it did several queries and read probably a dozen results before answering. And it correctly said that little information was available on the topic I asked about, which impressed me because I expect it to hallucinate in that situation instead.
I'm working on making it possible to feed LLMs live data from websites.
It is challenging because injecting webpages into LLM prompts directly has a mixed result, especially if there's a smidge of ambiguity or complexity in the webpage HTML, so part of my work is on automatic methods to economically prune, format and feed web data to LLM RAG pipelines.
You can follow my progress on twitter if you're interested (I tweet about other things, don't expect LLM/Data stuff only :)
Have you not been using chatGPT for programming questions this year? The implied criticism you're making became out of date when GPT4 was released. Of course, it makes things up when you stretch it, but the utility is no longer in question. What did you think all the fuss has been about?
Hasn't really been a problem. The way I use it is like
"Explain this concept to me."
"Give examples from the physical world and every day life."
"Elaborate on that last example."
"Give some examples from the domain of ___."
"Tell me about the history of the concept and why it was necessary to invent it."
"How does this concept relate to the concept of ___."
So I'm trying to approach what I want to learn from many different angles and build context and intuition. Sort of like talking to a tireless grad student tutor. Errors would become more obvious due to contradiction. I've caught it making some mistakes but they were sort of trivial and I could see the spirit of what it was getting at. Plus, "truth" tends to "click."
I just tried it out and I have to say I am blown away. I've never been someone who likes the voice capabilities (tried Siri once or twice, hated it) for just about any service as I can type pretty quickly but this is intelligent enough that I can have an unstructured conversation and still get to the results I want quickly. Very impressive.
I feel blown away and creeped out at the same time. It sounds so natural that's it's creepy. Only question I have now is that is audio being streamed or generated on device.
It is streamed. They also offer this as an API for developers [0].
I just wish they'd offer more languages as it currently is English only. It can speak German and Spanish, but with a very clear English accent, which makes it kind of funny and makes one think about why the accent sounds so real.
Just tried now, and it was pretty good. Some slight accent that I wasn't able to detect (Minas Gerais?). Similar / better than the artificial voice of other assistants in pt-br. I'm using Sky's voice, FWIW.
When I tried it right when it was released, it was some really heavy accent from Piracicaba. People there talk like that but ChatGPT seemed to be doing a forced impression of that.
It seems similar to what other commenters are saying about it taking in German with an American accent.
It's no big deal but Siri/Google/Alexa voices are better normalized, IMHO. I don't mind heavy accents at all in humans but they get annoying after a while for a voice assistants, especially if it's not my accent :)
I tried it now but it doesn't seem to be working for me. Maybe related to the outage yesterday (is it still going on?). It just listens but never replied.
I just got the Rayban Meta glasses an hour ago, put them on, said "Hey Meta" and I was interacting with an AI with my voice, and hearing a response back. It's so unbelievable and surreal I don't understand why Siri/Google assistant are not doing this already.
> I don't understand why Siri/Google assistant are not doing this already.
My theory is that voice assistants are made to be cheap entry points to those ecosystems. LLMs, on the other hand, are incredibly expensive to run. No GAFAM wants to pay dozens of cents when you ask their assistant to switch on the lights.
Well, Siri is not doing it because its „intelligence“ basically sounds to a bunch of if-then-else statements.
It’s still the most embarrassing Apple thing in existence. So very bad.
The best part about Siri is that what it replaced was insanely better at the specific task of reproducing music. Why does my iPhone in 2023 need to connect to the internet and shamefully fail when I try to play a song or even change the volume.
If anyone still remembers what we had before Siri, you’ll know how well it worked.
Interesting - this was actually a feature that got me to convert. I have some slight annoyances with how it handles natural pauses as I collect my thoughts and speak, but overall it's been great.
I've had some interesting discussions and it's helped me structure some thoughts by asking questions and follow-ups, then summarizing our conversation... all while I'm on a walk.
One fun thing I did with my family was have it do an interactive adventure story starring us. We had an adventure, and then used the built in DALL-E to generate images of scenes from our adventure.
> I have some slight annoyances with how it handles natural pauses as I collect my thoughts and speak
I hate this about all voice assistants. They force you to speak in an unnatural way.
It seems like there’s nothing intelligent about when they decide to respond, it’s like they just wait for X milliseconds of silence instead of using the context of what’s being said like a human would.
Sometimes it’s the opposite problem. You finish asking your question but you gotta wait for the assistant to pick it up whereas a human would understand that you’re finished talking based on what you said.
It might seem small and unimportant but I really think it’s one of the main reasons why voice assistants feel so… artificial.
I think the same thing can happen when speaking to anyone with different societal/cultural factors than yours. We just become more accustomed to it over time and it's less noticed. I think if this GPT had a big green alien face then we would find speaking to it less strange, somehow.
One of the very few things I like about Alexa is that it has an option for being more forgiving about stammers and pauses. It actually works pretty well.
Maybe the same will make it into ChatGPT at some point.
Yeah, I wish there was an option to verbally cue that you were finished talking... like an "over and out" thing. Do you know about the feature where you can press the circle to force it to listen and then release for it to answer?
It's interesting how in my frustration I intuitively tried that push-to-talk sorta-feature and it worked. Integrating something like Whisper and streaming live text could be neat, especially considering how different cultures handle conversational pauses and turn-taking. Wondering why there's no move towards full-duplex conversations in such tools.
While walking the dog today, it talked me through some trade-offs between DBSCAN and isolation forests. Walking + verbalizing the problem is a very different and positive experience for me.
I've also used it several times on ~15-20min drives to memorize something I wanted to have available for immediate recall. I had it chunk & quiz me, and by the end of the drive I had it down pat. Fun use of drive time.
A word of caution - I've asked ChatGPT 3.5 to generate quizzes based on books before and while most answers were right, a few were technically wrong, and some were outright fabrications (presented very confidently!)
On a computer, I'd paste in a corpus of some sort as a regular ChatGPT message - just because it's easier to accumulate a big string. Note with GPT4 Turbo and the recent UI upgrades, the context window is so large now that you can paste a sizable body of knowledge, possibly even as an attached file.
I'd then switch to the phone and retrieve the chat from History.
Here's an example prompt I just used to help my son prepare for a DMV written test:
```
I'm going to paste a large list of questions and answers and then switch to voice mode. Once I indicate that I'm ready, begin quizzing me on these questions. Feel free to rephrase slightly. My goal is to achieve complete retention of all of these questions through quizzing and spaced repetition. The questions are California DMV questions. I am preparing to take the written test.
1. *Q:* You may drive off of the paved roadway to pass another vehicle.
*A:* Under no circumstances.
2. *Q:* You are approaching a railroad crossing with no warning devices and are unable to see 400 feet down the tracks in one direction. The speed limit is...
*A:* 15 mph.
...
```
Wow I just tried the app with voice and the quality is amazing. It felt creepy. It is not a person I was talking to but the response time and quality is so good and I can't help myself but think I am talking to a real person. It is eerie.
Edit: Based on the response time, I don't understand where the audio is being generated, is it local?
I'm wondering which part of this link I should focus on: (1) ChatGPT Voice is now free for all users (2) Greg Brockman is retweeting an announcement from a company he left
Can somebody who knows their stuff about voice synthesis explain to me why chatGPT sounds like it's speaking German with an American accent when I speak to it in German? It's really surprising how believably it sounds like a native American-speaker speaking excellent, just not accent-free, German.
I would imagine that voice synthesis models would somehow be trained on data from native speakers, so why the accent?
My immediate assumption was that Chicago has a high Polish population, not necessarily that it's any different from other Polish American ones.
edit: It would appear so:
> Chicago is a city sprawling with Polish culture, billing itself as the largest Polish city outside of Poland, with approximately 185,000 Polish speakers, making Polish the third most spoken language in Chicago.
I am going to guess that 99% of the source data, models, training, correction, verification is translated by Americans.
Translating a language is difficult on top of regional dialects creates additional complications.
If someone is talking long enough to me I can identify their birth State based upon their American accent. Every State has a different pronunciation of certain words which can leak location data.
Guessing you're Californian, educated parents, about 34 years old. No siblings.
Not sure how you arrived at that guess, but generally I guess that describes the largest demographic of HN users.
Without AI but based on some freely available statistics combined with post history, we can say definitively (assuming honesty on OP’s part) that OP is 35 or 36, and was either born in Germany or moved there when they were young, and they likely still live there or at least have a strong affinity to Germany.
We can reasonably speculate that the fact that they speak German and likely live there, as well as the standard of their English drastically increases the likelihood their parents were educated. Germany’s current total fertility rate is 1.5. It was likely higher when OP was born back in 1986/7 but I couldn’t immediately find any data on that and didn’t look too hard. Given the fact that TFR is a population mean, this suggests a high proportion of single-child families, and generally the more educated one is, the fewer children, so it’s a decent guess that OP is an only child (apparently more than 50% of German families have only one child), but I don’t see anything that would make this a sure fire bet.
I was able to complete some further analysis which could reveal more likely truths about OP such as gender, political orientation, sexuality, etc., but I think this goes far enough without starting to doxx them.
I expect that the training data for the voice AI is overwhelmingly English. Native speakers of other languages will be a small minority. I'm sure it will improve over time.
It’s a good question. I’m not sure of exactly the answer, but I suspect the answer is similar to the answer “why do Americans speak German with an American accent?”
If you learn how to pronounce specific vowels, consonants, etc. in a particular way, it takes a LOT of effort to learn how to pronounce these in a different way. You can approximate to a good extent, but most researchers say that if you don’t develop this skill as a child, you won’t ever be able to pronounce things in a way that sounds like a native speaker.
Presumably, the models have been exposed to significantly more American accents than other accents, and learning how to pronounce phonemes with subtle differences without accepting a “close enough” approximation is a big challenge, especially given that there is already a threshold for acceptability at which level you can still sound like a native English speaker.
Am I the only one who doesn't like these specific voices? The quality is incredible, but they feel too cheery/enthusiastic/casual and it just gets annoying after a while.
I made an iOS shortcut a while ago that uses Siri with the ChatGPT app (it has iOS shortcut bindings) and despite Siri being a useless pile of junk compared to this, I actually prefer Siri's voice to this in some ways, because it doesn't feel so over the top.
Maybe this is partly because of different cultural expectations between the USA and Europe? Or maybe I'm just being too cynical and ChatGPT really is that happy talking with me!...
I don’t care about the voices themselves but the speech recognition is borderline unusable sometimes. It interjects when it shouldn’t and will frequently hear things incorrectly.
At one point it misinterpreted me mentioning “tai chi” as “I can’t breathe” and responded with advice about medical emergencies.
Do you mean Siri's voice recognition? If so, 100% agreed. My iOS shortcut uses OpenAI's Whisper API for voice recognition, and Siri (English United Kingdom - Siri Voice 1) for text to speech.
I really like dictating things sometimes, and Whisper is perfect for that (automatic paragraphs inside the model itself would be nice but not a big deal).
If anyone is interested - the "Whisper speech recognition in iOS" part is based on this shortcut I found that you can easily use yourself on both iOS and MacOS (free except for the OpenAI API usage fees obviously): https://giacomomelzi.com/transcribe-audio-messages-iphone-ai...
There are several versions of Whisper which have been distilled and can run locally, so I don’t see what advantage making API calls would be other than increased latency and decreased reliability and data security.
That's really interesting, Whisper is generally considered the current state of the art in STT and I've personally never experienced errors like the ones you describe. I've actually never had an error from Whisper.
First question, is there another STT you have used which works better for you?
Second question, is there any reason your voice might be considered unusual, like having a strong Welsh, Irish, or Indian accent, or being Deaf or Hard of Hearing?
Yeah, whisper is pretty good out of the box in my experience, but the vast majority of the time I’m using it in my car. So the conditions aren’t ideal, or are out of distribution for Whisper. However CarPlay is detectable and common enough from what I’ve heard.
Second, even if the transcription is correct, it cuts me off at inappropriate times. It’s hard to talk naturally without pauses.
Oh that's really interesting. Probably an acoustic environment it's not used to, like you said, but also people talk differently when they're driving. Like the cadence of our speech is significantly different because of the way our mental focus changes. I have to imagine that changes some things.
Of all of them, Sky's voice seems the most sober. I've been using it as a Plus subscriber for several weeks now and am also very impressed.
Yes, sometimes it thinks I'm done speaking when I'm not, but on the whole it's very good. Siri/Alexa, et al are not only unusable but are now supremely frustrating.
It is very much a cultural thing, the voice equivalent of decorating your Instagram. Ordering pizza after a 60h work week? Well, better make it sound like fun!!
Nope, you're not the only one. I posted as well: they sound to me like your classic, well-trained call center agents: Fake friendly (but please kill me now).
Reminds me way too much of some of the people I had to talk to, when cleaning up my mother's affairs. Places trying to get me to pay bills I did not owe, call center agents "cheerfully" following scripts that they themselves hated. The voices sound exactly like that.
Give me a neutral voice. This is a computer I'm talking to, not a fake friend.
It is probably cultural or linguistic. I love audio books but I cringe when I find a book I want to listen to that has an English voice actor. I don't think it is just the accent but all the pacing and emphasis.
I also though don't like most the chatGPT voice models besides for Sky. Sky to me is really good. Robertson Dean reading an audio book is perfection but Sky is pretty awesome.
I should add that as an American there are a ton of American voice actors that ruin books for me too. Sometimes this can be fixed if played at 1.2X speed.
I wonder why they launched it today, when it was supposed to be a week-off for all OAI employees and amidst whatever else is going on?
One possibility is to capture Thanksgiving gatherings as a growth hack where people demo this to their families/friends and increase app downloads for OAI.
Almost all of the OpenAI team signed a letter saying they would resign if the board is not replaced, so safe to say that they do care about the leadership drama; perhaps more than the rest of us.
There was at least one OpenAI employee on Blind [0] saying that they (unclear who "they" is but presumably some other employees) pressured employees into signing and called him in the middle of the night to do it. I just googled it and found a thread as well [1].
My theory is that it was nearing ready for release so they pushed it out to deflect some of the bad press over recent drama. I'd suspect the challenge with releasing something like this is about 2% "the code" and 98% scaling it out to everyone and making that not explode.
Is there a way to paste text into the mobile app and get the voice to read it? I can't for the life of me figure out how to do this. There's no text input shown while in the voice feature despite ChatGPT telling me to "just type what you want me to read"
I'm not sure if there's a better way but I've typed out requests via text, and as part of the request I type "read the results to me when I ask for them" or something along those lines. Then I turn on the voice mode and make the request.
I don't disagree but I'm just stating what is available in the app. Voice or text, they treat chats as a single thing.
I understand the downvotes for showing disapproval at OpenAI's policies (or big tech's), but I don't know what else could I say. Should I have just replied with a snarky comment saying "a-ha! you expect them to respect your privacy? pffff"
The setting is only saved client side which is a very odd design decision. I guess they made it that way so that if you disable the flag at one point (yes, it's active by default), the change only effects future requests. This whole privacy situation with openai makes me really angry with Sam Altman, who seems not to care about that at all - this setting seems to be specifically designed to be privacy-disrespecting. It reminds me on the early days of facebook when they set all your profile information to be public, but openai is much worse, because the information you put into the prompt is likely much more sensitive.
What bothers me about HN currently is that the majority seems to not be bothered by it, stating that OpenAI tries to be ethical, that Sam can be trusted etc.
Has everyone forgotten about Google? Facebook? Every other SV Tech-Startup? Why would OpenAI be different?
"They're a non-profit". Yeah, sure. As if Wall Street, M$ et al would just not try to monopolize and commercialize AI as soon as possible.
302 comments
[ 3.0 ms ] story [ 284 ms ] threadhttps://twitter.com/OpenAI/status/1727065166188274145
Changed from https://twitter.com/gdb/status/1727067288740970877 above, plus I degregged the title. I'm sure he won't mind. (submitted title was "ChatGPT Voice for All Free Users Announced (By Greg Brockman)")
Thanks!
It's suggesting to ask it where to order 195 pizzas from.
Does it have access to active business records like that?
You can follow my progress on twitter if you're interested (I tweet about other things, don't expect LLM/Data stuff only :)
https://twitter.com/omarkamali
https://twitter.com/themonitoro
No, I'm a fighter pilot in EVE Online.
"Explain this concept to me."
"Give examples from the physical world and every day life."
"Elaborate on that last example."
"Give some examples from the domain of ___."
"Tell me about the history of the concept and why it was necessary to invent it."
"How does this concept relate to the concept of ___."
So I'm trying to approach what I want to learn from many different angles and build context and intuition. Sort of like talking to a tireless grad student tutor. Errors would become more obvious due to contradiction. I've caught it making some mistakes but they were sort of trivial and I could see the spirit of what it was getting at. Plus, "truth" tends to "click."
I just wish they'd offer more languages as it currently is English only. It can speak German and Spanish, but with a very clear English accent, which makes it kind of funny and makes one think about why the accent sounds so real.
[0] https://platform.openai.com/docs/guides/text-to-speech
It seems similar to what other commenters are saying about it taking in German with an American accent.
It's no big deal but Siri/Google/Alexa voices are better normalized, IMHO. I don't mind heavy accents at all in humans but they get annoying after a while for a voice assistants, especially if it's not my accent :)
I tried it now but it doesn't seem to be working for me. Maybe related to the outage yesterday (is it still going on?). It just listens but never replied.
Among others...
Had to check myself.
> Afrikaans, Arabic, Armenian, Azerbaijani, Belarusian, Bosnian, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Marathi, Maori, Nepali, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tagalog, Tamil, Thai, Turkish, Ukrainian, Urdu, Vietnamese, and Welsh.
The English pronunciation is perfect, but some Russian words do not have the correct pronounciation or emphasis.
I'm wondering if I can train the model to pronounce these words more accurately.
I really feel like I live in the future.
My theory is that voice assistants are made to be cheap entry points to those ecosystems. LLMs, on the other hand, are incredibly expensive to run. No GAFAM wants to pay dozens of cents when you ask their assistant to switch on the lights.
If anyone still remembers what we had before Siri, you’ll know how well it worked.
I've had some interesting discussions and it's helped me structure some thoughts by asking questions and follow-ups, then summarizing our conversation... all while I'm on a walk.
One fun thing I did with my family was have it do an interactive adventure story starring us. We had an adventure, and then used the built in DALL-E to generate images of scenes from our adventure.
I hate this about all voice assistants. They force you to speak in an unnatural way.
It seems like there’s nothing intelligent about when they decide to respond, it’s like they just wait for X milliseconds of silence instead of using the context of what’s being said like a human would.
Sometimes it’s the opposite problem. You finish asking your question but you gotta wait for the assistant to pick it up whereas a human would understand that you’re finished talking based on what you said.
It might seem small and unimportant but I really think it’s one of the main reasons why voice assistants feel so… artificial.
That and long ass responses.
I think the same thing can happen when speaking to anyone with different societal/cultural factors than yours. We just become more accustomed to it over time and it's less noticed. I think if this GPT had a big green alien face then we would find speaking to it less strange, somehow.
Maybe the same will make it into ChatGPT at some point.
It would be great to have an option to have ChatGPT stop when saying "over" - it does not feel natural to have to blurt the questions all at once.
you can ask it to be terse and shorten answers in the custom instructions section.
I've also used it several times on ~15-20min drives to memorize something I wanted to have available for immediate recall. I had it chunk & quiz me, and by the end of the drive I had it down pat. Fun use of drive time.
Can you go a bit into how this works? How do you prompt it? This is the first use of ChatGPT I've heard of that would directly benefit me.
I'd then switch to the phone and retrieve the chat from History.
Here's an example prompt I just used to help my son prepare for a DMV written test:
``` I'm going to paste a large list of questions and answers and then switch to voice mode. Once I indicate that I'm ready, begin quizzing me on these questions. Feel free to rephrase slightly. My goal is to achieve complete retention of all of these questions through quizzing and spaced repetition. The questions are California DMV questions. I am preparing to take the written test. 1. *Q:* You may drive off of the paved roadway to pass another vehicle. *A:* Under no circumstances.
2. *Q:* You are approaching a railroad crossing with no warning devices and are unable to see 400 feet down the tracks in one direction. The speed limit is... *A:* 15 mph. ... ```
Edit: Based on the response time, I don't understand where the audio is being generated, is it local?
Also this Call Annie requires Google sign in.
https://news.ycombinator.com/item?id=38374582
I assume there will be a MS ChatGPT if OpenAI folds though. Or just Bing.
Oops! Our systems are a bit busy at the moment, please take a break and try again soon.
I would imagine that voice synthesis models would somehow be trained on data from native speakers, so why the accent?
edit: It would appear so:
> Chicago is a city sprawling with Polish culture, billing itself as the largest Polish city outside of Poland, with approximately 185,000 Polish speakers, making Polish the third most spoken language in Chicago.
Translating a language is difficult on top of regional dialects creates additional complications.
If someone is talking long enough to me I can identify their birth State based upon their American accent. Every State has a different pronunciation of certain words which can leak location data.
Guessing you're Californian, educated parents, about 34 years old. No siblings.
Without AI but based on some freely available statistics combined with post history, we can say definitively (assuming honesty on OP’s part) that OP is 35 or 36, and was either born in Germany or moved there when they were young, and they likely still live there or at least have a strong affinity to Germany.
We can reasonably speculate that the fact that they speak German and likely live there, as well as the standard of their English drastically increases the likelihood their parents were educated. Germany’s current total fertility rate is 1.5. It was likely higher when OP was born back in 1986/7 but I couldn’t immediately find any data on that and didn’t look too hard. Given the fact that TFR is a population mean, this suggests a high proportion of single-child families, and generally the more educated one is, the fewer children, so it’s a decent guess that OP is an only child (apparently more than 50% of German families have only one child), but I don’t see anything that would make this a sure fire bet.
I was able to complete some further analysis which could reveal more likely truths about OP such as gender, political orientation, sexuality, etc., but I think this goes far enough without starting to doxx them.
you wouldn't believe it, but models haven't been trained yet. as usual.
If you learn how to pronounce specific vowels, consonants, etc. in a particular way, it takes a LOT of effort to learn how to pronounce these in a different way. You can approximate to a good extent, but most researchers say that if you don’t develop this skill as a child, you won’t ever be able to pronounce things in a way that sounds like a native speaker.
Presumably, the models have been exposed to significantly more American accents than other accents, and learning how to pronounce phonemes with subtle differences without accepting a “close enough” approximation is a big challenge, especially given that there is already a threshold for acceptability at which level you can still sound like a native English speaker.
I made an iOS shortcut a while ago that uses Siri with the ChatGPT app (it has iOS shortcut bindings) and despite Siri being a useless pile of junk compared to this, I actually prefer Siri's voice to this in some ways, because it doesn't feel so over the top.
Maybe this is partly because of different cultural expectations between the USA and Europe? Or maybe I'm just being too cynical and ChatGPT really is that happy talking with me!...
At one point it misinterpreted me mentioning “tai chi” as “I can’t breathe” and responded with advice about medical emergencies.
I really like dictating things sometimes, and Whisper is perfect for that (automatic paragraphs inside the model itself would be nice but not a big deal).
If anyone is interested - the "Whisper speech recognition in iOS" part is based on this shortcut I found that you can easily use yourself on both iOS and MacOS (free except for the OpenAI API usage fees obviously): https://giacomomelzi.com/transcribe-audio-messages-iphone-ai...
> free except for the OpenAI API usage fees
There are several versions of Whisper which have been distilled and can run locally, so I don’t see what advantage making API calls would be other than increased latency and decreased reliability and data security.
First question, is there another STT you have used which works better for you?
Second question, is there any reason your voice might be considered unusual, like having a strong Welsh, Irish, or Indian accent, or being Deaf or Hard of Hearing?
Second, even if the transcription is correct, it cuts me off at inappropriate times. It’s hard to talk naturally without pauses.
I haven’t used a better transcription model, no.
Yes, sometimes it thinks I'm done speaking when I'm not, but on the whole it's very good. Siri/Alexa, et al are not only unusable but are now supremely frustrating.
Reminds me way too much of some of the people I had to talk to, when cleaning up my mother's affairs. Places trying to get me to pay bills I did not owe, call center agents "cheerfully" following scripts that they themselves hated. The voices sound exactly like that.
Give me a neutral voice. This is a computer I'm talking to, not a fake friend.
I also though don't like most the chatGPT voice models besides for Sky. Sky to me is really good. Robertson Dean reading an audio book is perfection but Sky is pretty awesome.
I should add that as an American there are a ton of American voice actors that ruin books for me too. Sometimes this can be fixed if played at 1.2X speed.
One possibility is to capture Thanksgiving gatherings as a growth hack where people demo this to their families/friends and increase app downloads for OAI.
[0] https://teamblind.com
[1] https://www.teamblind.com/post/OpenAI-employees-did-you-sign...
I understand the downvotes for showing disapproval at OpenAI's policies (or big tech's), but I don't know what else could I say. Should I have just replied with a snarky comment saying "a-ha! you expect them to respect your privacy? pffff"
Has everyone forgotten about Google? Facebook? Every other SV Tech-Startup? Why would OpenAI be different?
"They're a non-profit". Yeah, sure. As if Wall Street, M$ et al would just not try to monopolize and commercialize AI as soon as possible.