OpenAI语音助理的回应有多厉害？几乎辗轧Siri和Alexa-纽约情报站

HILOVEYOUTU 发表于 2024-5-13 21:36:47

OpenAI语音助理的回应有多厉害？几乎辗轧Siri和Alexa

<picture style="box-sizing: border-box;"> <img alt="OpenAI发表了新的GPT-4o模型，含有语音助理新功能，实时响应的速度和内容令人惊艳。路透" class=" lazyloaded" data-src="https://pgw.worldjournal.com/gw/photo.php?u=https://uc.udn.com.tw/photo/wj/realtime/2024/05/14/29655174.jpg&x=0&y=0&sw=0&sh=0&sl=W&fw=800&exp=3600&q=75" src="https://pgw.worldjournal.com/gw/photo.php?u=https://uc.udn.com.tw/photo/wj/realtime/2024/05/14/29655174.jpg&x=0&y=0&sw=0&sh=0&sl=W&fw=800&exp=3600&q=75" _mstalt="464117706" _msthash="192" style="box-sizing: border-box; border-style: none; object-fit: contain; width: 800px; height: 533px;"></picture><figcaption _msttexthash="453645348" _msthash="193" style="box-sizing: border-box; line-height: 1.4; padding-top: 10px; padding-bottom: 10px; color: rgb(126, 126, 126); background-image: initial; background-position: initial; background-size: initial; background-repeat: initial; background-attachment: initial; background-origin: initial; background-clip: initial;">OpenAI发表了新的GPT-4o模型，含有语音助理新功能，实时响应的速度和内容令人惊艳。路透</figcaption><span _istranslated="1" style="box-sizing: border-box; background-color: initial; color: rgb(126, 126, 126); text-decoration-line: none; border-bottom: 1px solid rgb(126, 126, 126); transition: color 0.3s linear 0s, border 0.3s linear 0s;">OpenAI发表了新的GPT-4o模型，含有语音助理新功能，实时响应的速度和内容令人惊艳，除了能像人类在正常对话中一样快速响应、即兴唱歌、能嘲笑不好笑的笑话或者自嘲，还能在相机镜头协助下，帮忙人类解决写在课本上的数学问题。 <div class="video-container" style="box-sizing: border-box; position: relative; padding-bottom: 450px; padding-top: 30px; height: 0px; overflow: hidden; margin-bottom: 20px; color: rgb(0, 0, 0);"><iframe width="640" height="360" src="https://www.youtube.com/embed/vgYi3Wr7v_g?enablejsapi=1&origin=https%3A%2F%2Fwww.worldjournal.com" frameborder="0" allowfullscreen="" data-gtm-yt-inspected-6="true" id="862132441" data-gtm-yt-inspected-30403861_28="true" title="Say hello to GPT-4o" style="box-sizing: border-box; position: absolute; top: 0px; left: 0px; width: 800px; height: 480px; max-width: 100%;"></iframe></div>OpenAI在YouTube频道上张贴了16段影片，展示GPT-4o模型语音助理与人类的各种聊天互动，像是利用相机判断用户脸上的笑容表情、推测有快乐心情，判断用户穿了什么颜色的衣服与身处的空间环境，藉此展开聊天话题。当用户提到，今天有个发布会要介绍GPT-4o的语音助理时，「她」瞬间惊讶并害羞笑出来，回应说「我？是要介绍我？」根据OpenAI说法，GPT-4o模型对语音输入的反应最快可达0.232秒，实际上平均只要大约0.32秒就可响应，号称与人类对话时的速度差不多。这些视频中，除了大秀实时翻译功能，展示者还在纸上写了一道代数问题「3 X+1=4」询问ChatGPT，语音助理利用相机镜头解读算式之后，一步一步教导使用者解出答案。<div class="video-container" style="box-sizing: border-box; position: relative; padding-bottom: 450px; padding-top: 30px; height: 0px; overflow: hidden; margin-bottom: 20px; color: rgb(0, 0, 0);"><iframe width="640" height="360" src="https://www.youtube.com/embed/RI-BxtCx32s?enablejsapi=1&origin=https%3A%2F%2Fwww.worldjournal.com" frameborder="0" allowfullscreen="" data-gtm-yt-inspected-6="true" data-gtm-yt-inspected-30403861_28="true" id="168880253" title="Live demo of GPT-4o vision capabilities" style="box-sizing: border-box; position: absolute; top: 0px; left: 0px; width: 800px; height: 480px; max-width: 100%;"></iframe></div>另一段视频显示用户要求ChatGPT讲一段床边故事「机器人与爱」，在下指令希望语音助理的语调更具戏剧性后，ChatGPT立刻「变声」，用演话剧的方式讲故事，甚至马上切换成机器人说话的语调，最后甚至唱了一段歌为故事画下句点。<div class="video-container" style="box-sizing: border-box; position: relative; padding-bottom: 450px; padding-top: 30px; height: 0px; overflow: hidden; margin-bottom: 20px; color: rgb(0, 0, 0);"><iframe width="640" height="360" src="https://www.youtube.com/embed/D9byh4MAsUQ?enablejsapi=1&origin=https%3A%2F%2Fwww.worldjournal.com" frameborder="0" allowfullscreen="" data-gtm-yt-inspected-6="true" data-gtm-yt-inspected-30403861_28="true" id="113527169" title="Live demo of GPT4-o voice variation" style="box-sizing: border-box; position: absolute; top: 0px; left: 0px; width: 800px; height: 480px; max-width: 100%;"></iframe></div>当然，GPT-4o模型同样善于协助分析和撰写代码，也能分析图表，而且这回还是用语音生动地与用户对话。<div class="video-container" style="box-sizing: border-box; position: relative; padding-bottom: 450px; padding-top: 30px; height: 0px; overflow: hidden; margin-bottom: 20px; color: rgb(0, 0, 0);"><iframe width="640" height="360" src="https://www.youtube.com/embed/mzdvw_euKlk?enablejsapi=1&origin=https%3A%2F%2Fwww.worldjournal.com" frameborder="0" allowfullscreen="" data-gtm-yt-inspected-6="true" data-gtm-yt-inspected-30403861_28="true" id="516929948" title="Live demo of GPT-4o coding assistant and desktop app" style="box-sizing: border-box; position: absolute; top: 0px; left: 0px; width: 800px; height: 480px; max-width: 100%;"></iframe></div>另一段视频还显示，将两台使用GPT-4o模型的手机放在旁边时，「两位」语音助理甚至能够互相对话聊天、即兴唱歌。<div class="video-container" style="box-sizing: border-box; position: relative; padding-bottom: 450px; padding-top: 30px; height: 0px; overflow: hidden; margin-bottom: 20px; color: rgb(0, 0, 0);"><iframe width="640" height="360" src="https://www.youtube.com/embed/MirzFk_DSiI?enablejsapi=1&origin=https%3A%2F%2Fwww.worldjournal.com" frameborder="0" allowfullscreen="" data-gtm-yt-inspected-6="true" data-gtm-yt-inspected-30403861_28="true" id="355289193" title="Two GPT-4os interacting and singing" style="box-sizing: border-box; position: absolute; top: 0px; left: 0px; width: 800px; height: 480px; max-width: 100%;"></iframe></div>在其他影片中，GPT-4o模型还能嘲笑笑话、正确识别猜拳的获胜者，也能根据蛋糕插了蜡烛来判断这可能是某人生日，还唱了一首生日快乐歌。对于这场发表活动，MarketWatch网站发表一篇文章，题为「OpenAI是否刚推出了Siri杀手？」有网友盛赞这把ChatGPT功能提升到了全新水平、未来是如此光明。<span _istranslated="1" style="box-sizing: border-box; background-color: initial; color: rgb(126, 126, 126); text-decoration-line: none; border-bottom: 1px solid rgb(126, 126, 126); transition: color 0.3s linear 0s, border 0.3s linear 0s;">英伟达高级研究科学家Jim Fan则指出，OpenAI的语音助理「生动逼真而且甚至有点轻浮」，还让人回想起2013年的科幻电影「云端情人」。<div align="left"><div align="center">点击下面文字可快速查看或发布对应的便民信息！</div><div align="center">纽约情报站让您的生活变的更简单</div><div align="center"><hr class="l" style="clear: both;"></div></div><div align="left"><div align="center">:hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=62&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">电召车</a> :hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=5&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">顺风拼车</a></div><div align="center"><hr class="l" style="clear: both;"></div></div><div align="left"><div align="center">:hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=21&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">汽车买卖</a> :hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=64&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">便捷搬家</a></div><div align="center"><hr class="l" style="clear: both;"></div></div><div align="left"><div align="center">:hug:<a href="https://bbs.nyinfor.com/plugin.php?id=fn_job&formapp=&footernav=true&idu=1" target="_blank">招聘求职</a> :hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=92&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">店铺转让</a></div><div align="center"><hr class="l" style="clear: both;"></div></div><div align="left"><div align="center">:hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hb&ac=cat&cat_id=88&province=&city=&dist=&orderby=&lat=&lng=&filter=" target="_blank">房屋出售</a> :hug:<a href="https://bbs.nyinfor.com/plugin.php?id=xigua_hs&idu=1" target="_blank">商家黄页</a></div></div><div align="left"><div align="center">简介：纽约情报站是汇集全美75万粉丝的公众平台。除了实时新闻、找工招工信息发布、app社区互动，更有微信公众号推文探店等。如果你投稿、爆料、活动策划、商务合作，或者想邀请我们去探店，请联系主编微信： nyinfor</div></div>

墨尘发表于 2024-5-13 23:15:15

不错，厉害

页: [1]

纽约情报站's Archiver

OpenAI语音助理的回应有多厉害？ 几乎辗轧Siri和Alexa

OpenAI语音助理的回应有多厉害？几乎辗轧Siri和Alexa