展示HN：自定义播客摘要工具
这个工具允许用户根据自己的偏好对任何播客进行总结。随着时间的推移，应用程序通过自定义标签和人工智能聊天功能了解您想要提取的信息。每次互动都会记录您实际想从播客中获取的信息，这有助于更好地定制您的总结。
我之所以开发这个工具，是因为我听了很多超过一个小时的播客，但往往没有时间全部听完，结果导致我有一个永无止境的待听列表，想听却总是抽不出时间。由于我感兴趣的许多播客内容，等我听的时候信息可能已经过时。
我意识到我有两个选择：一是以2倍速听播客，同时处理我的日常工作；二是承认我偶尔会错过一些信息。
后来我意识到，我可能不需要听一两个小时就能获取我想要的信息。对我而言，相关的信息可能只占大多数播客内容的25%以下，但通用的总结工具未必有效，因为我无法保证我想要的见解会被包含在总结中。
另外，我可以找到播客的文字稿，然后将其输入到大型语言模型（LLM）中并进行提示，但这可能会有些耗时。
因此，我将所有功能整合到一个应用中。现在，我可以在不到10分钟的时间内从六个小时的播客中获取我所需的信息。
如何使用：1. 访问 getdecant.app/try 并输入密码“showhn”；2. 在搜索框中搜索任何播客。一旦找到您要找的播客，点击“添加”；3. 如果您提前知道想要了解的特定信息，请点击“+ 新标签”，并写下您想知道的内容并给它一个标题。例如，我最喜欢的播客之一是《How I Built This》，所以我可能会创建一个名为“商业模式”的标签，内容为“提供本集讨论的商业的分销和收入模型的详细信息。”；4. 点击您添加的播客的缩略图；5. 如果您创建了自定义标签，请点击该标签以激活它；6. 选择您想要总结的集数，这将带您到播放器页面，您可以在此收听该集；7. 点击“生成总结”；8. 使用聊天功能深入了解更多细节或询问总结中未包含的内容。
祝您使用愉快！期待您的反馈。
查看原文
This tool allows users to summarize any podcast according to their own preferences. Over time, the app learns about what information you want to extract via custom tags and an AI chat feature. Each interaction logs news information about what information you actually want from your podcasts which leads to better customization of your summaries.
I built this because I listen to a lot of podcasts that are over an hour long but I often don't have time to listen to them all and I end up with a never ending backlog of episodes that I want to listen to but can never get around to. Due to the content of many of the podcasts I'm interested in, information can be outdated by the time I get around to listening.<p>I figured I had two choices, listen at 2x speed while multi-tasking my day job, or just concede that I was going to have to miss some information once in a while.<p>Then I realized that I probably don't need to listen for an hour or two to get the information I want. The information relevant to me probably constituted less than 25% of most podcasts but generic summarizers won't necessarily work because I can't guarantee that the insights I want will be included in the summary.<p>Alternatively, I could find the transcript for a podcast and feed it to an LLM and prompt it, but that can be somewhat time consuming.<p>So I built everything into one app. Now I can get exactly the information I need from six hours worth of episodes in under 10 mins.<p>HOW IT WORKS: 1. Go to getdecant.app/try and enter the password "showhn" 2. Search for any podcast in the search field. Once you find the one you're looking for, click "Add" 3. If you know ahead of time that there will be specific information you want to know click "+ New Tag" and just write in what you want to know and give it a title. For example, one of my favorite podcasts is How I Built This, so I might create a tag called "Business Models" that reads, "Provide the details on the distribution and revenue model of the business discussed in this episode." 4. Click the thumbnail image of the podcast you added. 5. If you created a custom tag, click that tag to activate it for that podcast. 6. Select the episode you want summarized, it will take you to the player page where you can listen to the episode. 7. Click "Generate Summary" 8. Use the chat feature to drill into more detail or to ask about something that's not in the summary.<p>Enjoy! Looking forward to feedback.