š Simple solution to your large audio transcription woes šļø Split any large audio file *directly* in your browser š” Transcribe the partitions with OpenAI's Whisper š Get a complete transcript with just one click ā No more size limits. No more manual effort
Hi, Product Hunt community,
My name's Trevor Healy, I'm a frontend engineer. I'm excited to share my first Product Hunt creation "Whisper for Large Audio" (product naming is my strong suit) with you today!
https://whisper.trevorhealy.me
As some of you may know, Whisper is an audio transcription technology created by OpenAI. In my experience, it's a solid balance of cost-effectiveness, accuracy, and ease of use. However, I stumbled upon a constraint when I began transcribing over 100 hours of audio - the 25MB upper file size limit. For someone looking at transcribing days worth of audio content, sometimes several GB in size, this posed a challenge.
Sure, I could partition all the files locally, but what about the future audio files? Would this always be a manual process? What if someone non-technical wanted to achieve the same thing? I wondered: what if all you needed was the audio file and a browser? I decided to build a tool that anyone could use.
Introducing "Whisper for Large Audio":
šļø Utilizes FFmpeg in the browser using Web Assembly ā in my opinion, this is the coolest part of the product. This lets us split even the heftiest audio files right where you are: in your browser. This processing is memory intensive, and I notice it has spotty success on mobile devices.
š Transcribes each section using OpenAI's Whisper. Rather than send 100 partitions at a time (which will certainly be rate-limited), the app will concurrently process the partitions in batches. Since Whisper prices by source audio time, the 4 hour file cut into 100 pieces will be the same price as the large one source file.
š¤ No time-saving utility app would be complete without a final copy to save button that ties it all together. Simply bundles the partitions into one large cohesive transcript.
The idea is simple: skip the need for servers to process the audio, eliminate manual slicing, one-click processing, and one-click copying. All within your browser.
I genuinely hope this tool is helpful to even one person. I love talking about these technologies and am keen to hear your thoughts, feedback, and ways you might find "Whisper for Large Audio" useful.
Cheers to making things a bit simpler! šāāļø
Love it. This is what actually makes Whisper immediately helpful for me @trevorwhealy. Super impressed that the partitioning happens directly in the browser.
Love it. This is what actually makes Whisper immediately helpful for me @trevorwhealy. Super impressed that the partitioning happens directly in the browser.
Whisper for Large Audio
Whisper for Large Audio
Portkey
Whisper for Large Audio