ChatGPT simply received a nifty new Superior Voice Mode earlier this week, and though it’s solely rolling out to a small subset of paying subscribers proper now – in alpha testing – we’ve now been handled to varied tasters of the characteristic in motion.
These are popping up on-line, on the likes of YouTube and X, with the fortunate, chosen ChatGPT Plus customers who’ve entry to the characteristic exhibiting it off throughout a variety of various duties. As The Verge studies, these embody requests to sing a music in a sure manner, or imitate accents, by means of to tackling the nuances of right pronunciation in languages.
In case you recall, this performance was really revealed on the GPT-4o launch a couple of months again. Nonetheless, Superior Voice Mode was delayed over obvious considerations round tightening up security with the characteristic, but it surely’s now right here, and really undoubtedly in motion as talked about – with some spectacular outcomes besides.
For instance, The Verge factors out ChatGPT giving a lesson within the pronunciation of French phrases to a consumer on YouTube, the place the AI is fairly useful.
Right here’s one other instance: a request to sing ‘Completely satisfied Birthday’ in a ‘soulful blues’ fashion. Or how about ChatGPT telling some jokes in distinction voices (shy, indignant)?
ChatGPT Superior Voice Mode counting as quick as it will probably to 10, then to 50 (this blew my thoughts – it stopped to catch its breath like a human would) pic.twitter.com/oZMCPO5RPhJuly 31, 2024
Lastly, try the above and beneath posts on X of ChatGPT’s Superior Voice Mode counting quick, after which tackling regional US accents.
ChatGPT Superior Voice Mode making an attempt numerous US regional accents pic.twitter.com/UvDeQUNHLpJuly 31, 2024
In case you’re eager to get in on the motion your self, we’ve been informed by OpenAI that each one ChatGPT Plus subscribers will get Superior Voice Mode later this yr. The complete rollout ought to be accomplished by the ‘finish of fall’ so everybody ought to have it by the point December will get right here, in concept.
Evaluation: 50 shades of cool
In case you’ve checked out the above demos – fairly cool, huh? If not, get checking…
There’s some critical consideration to element exhibited when it comes to making the Superior Voice Mode appear extra human-like and actual – notice the self-imposed synthetic degree of issue included into counting to 50 super-quickly, together with a pause for breath, a very neat contact.
Or the blues singing tour, which isn’t simply in regards to the precise singing – which is properly carried out, for positive – however the in-depth explanations of how the singer may method the music, and pure fashion and supply of the AI voice right here (and elsewhere). These AI interactions are pushed to new heights of realism right here, even when there are nonetheless wrinkles to be addressed.
By way of the latter, we weren’t so impressed with the US accents – although this was an enormous outdated ask, they usually have been a little bit higher when the consumer requested ChatGPT to emphasise them extra. And whereas the AI responses are usually very fast and to the purpose – and fluid – there’s the odd second of silence and confusion to be witnessed, when viewing a variety of those clips on-line.
Keep in mind, although, Superior Voice Mode continues to be in alpha, and on condition that, it’s actually fairly spectacular – strikingly good in some situations. This might be one of many areas through which AI strikes so quick, that it turns into scary…
You may additionally like…
