Show HN: Vocal Slice – Cut audio by selecting text, fully on-device
vocalslice.comI make audio tools, and this one came out of the clip-pulling loop: scrub to find the right line, set the in point, set the out point, name the file, repeat.
Vocal Slice transcribes your audio, then you can simply highlight the passage you want and the waveform jumps to exactly that span. Fine-tune with the handles, export a named clip.
Windows and macOS.
https://vocalslice.com wesley@vocalslice.com
Happy to answer anything in replies, please get in touch!
How do GitHub's terms and conditions like using GitHub Releases as storage for proprietary binaries? I noticed the download link points to https://github.com/terranivium/vocal-slice-releases ... and the license there https://github.com/terranivium/vocal-slice-releases/blob/mai... points at a "THIRD-PARTY-NOTICES.md" file that doesn't exist.
IMO, this is vibeable with tauri, with some reverse engineering of this. Worth a shot to oneshot these, fable and opus would chew through this and give a FOSS project I think in under 20$ token cost.
the core part of this is understanding the needs of audio professionals and their workflow.
the notices are available within the app, i'll make a note that these should also be publicly listed on the repo
If you extended this to allow assembling audio from multiple takes of the same dialogue, you’d have something very useful for audio drama post-production.
Thanks tene, from your comment i believe this functionality does exist, and i realise that i don't express it very clearly on my website. Essentially when you select text, it tells you how many 'matches' that text has. So if the same line is repeated, you can quickly seek between the matches to cut/AB multiple takes very quickly!
Great! But is the focus exporting short clips? I’m thinking more in terms of assembling long pieces of audio scene by scene, like 30+ mins. That’s a major workflow headache you could make radically easier. Depends on your target audience but you’d save them dozens of hours.
the initial idea came from exporting lines and multiple takes for video game voice over delivery, but the functionality is there for doing longer clips too by making longer text selections. i'd love to get some feedback from this group if you'd like to discuss further? wesley@vocalslice.com
Why is an annual license instead of one-time only purchase? Do you plan on adding additional features in the future that would bloat this project or otherwise justify perpetual payment?
great question - I had an older iteration of this project which was a much smaller scoped utility (without the waveform functionality) that was a one time purchase on itch. The idea with this project is to add additional features over time (um and ah removal, the ability to actually remove parts of the transcription and output audio).
I want to add, because I see this here almost every day, that making something a subscription is totally fine, regardless of updates.
I can offer something for free that costs me money, or I can charge money for something that costs me nothing.
"If I pay money for this, I want someone to suffer for an amount of time proportional to the amount and frequency paid" is not a healthy or helpful attitude in my opinion. (It is, however, the basis of most employment, which explains the constant confusion on this matter.)
Anyway, just want to say that you don't need to change your business model to satisfy people who complain on forums, and you don't need to justify the price or business model with your work or time or suffering. The value of the product is sufficient!
Really appreciate your thoughts on this, i totally agree - it is food for thought and as i monitor things it may make sense to tweak my approach. Bloating the tool would be counter productive for the sake of validating its cost, the idea long term is to improve, support and develop smart. Any financial model's goal in my eyes should be the means to allow people access, while justifying the product's existence.
Thank you for the amazing support so far, a few of you have reached out via email to share use cases I hadn't thought of. This tool was originally conceived around long voice acting recording sessions which were a pain to sift through, but the privacy aspect has implications in legal and NDA scenarios too.
Is it open source ? Curious if there is a breakdown on how it works. Looks really cool
currently no, the app isn't open source - however the app itself announces and is quite transparent on how it functions. it takes your audio, transcribes it locally using speech to text, recording the time stamps of each word. when you select text to make slices, Vocal Slice makes use of the time stamps to align the text selection with the correct audio in and out points, matching the text you have selected.
How is this different from descript? It seems like descript, but you took out the video functionality?
Descript is a timeline editor, so you delete text and the audio goes with it, and you come out with one finished piece. Vocal Slice goes the other way. You highlight a phrase and that span exports as its own file. Vocal Slice is also local rather than cloud based like Descript.
Cool idea. But I wonder if the text to speech is 100% accurate when that step takes place
Thank you! speech to text models can be extremely accurate, but it really depends on the audio you feed it. the beauty here with Vocal Slice is that it doesn't need to be perfect because the user is given the tools to fine-tine the selection via the waveform controls and listening to playback. over time i would like to solve the purely automated workflow though, and make the manual step less necessary.
I’d like to understand how this works.
This account was created only 13 days ago.
It has made only one submission, which is this post, and had only two comments before submitting it.
How does this happen?
I have an HN account that is 10 years old, but when I submit something, it often gets flagged almost immediately. So how exactly does HN’s flagging and ranking system work? Does account age or karma matter at all?
For reference:
user: terranivium created: 13 days ago karma: 18 about: vocalslice.com
flagging is entirely driven by user's flagging a post
Which part of this exactly looks like slop to you?
There are lots of AI-sloppy tells, starting from, let's see:
* the web site's layout (eyebrows, glows, em-dashes, middots, that certain breathless tone in the pricing box) (edit: see the source for https://vocalslice.com too)
* the changelog listing, eg. the fact that you can search by language code as a headline feature
* in the app, things like "Click here to choose an audio file to transcribe – swap it any time" are Claude-like copy; I'd be surprised if you _couldn't_ change the audio file at any time?
* looking at https://vocalslice.com/demo.webp, another claudeism is the glowy status indicator blob, it really likes to add that by default.
If OP gets a business going with this, more power to them and all the best, but it really is something that's buildable pretty easily with transformers.js and models like onnx-community/whisper-small_timestamped. (Speaking of which, I hope OP is hosting the models somewhere themselves and not trusting, say, Hugging Face with the continuity of their business.)
the truly important thing from discussion with users is the actual workflow optimisation that the app provides, but i totally agree some of the copy on the website could use improvement and the comments on this show hn have made it clear that i'm not communicating things as well as i could
UX-wise, from what I can glean from the .webp, I'd personally (having worked in this same space) appreciate being able to see the slices I already have without having to switch tabs back and forth, for one...
great suggestion - the transcription tab is already quite limited for real estate. i recommend trying out the free trial to get a feel for the workflow.
You might also want to take HH feedback with a grain of salt, it's a very particular demographic.
Classic example:
https://news.ycombinator.com/item?id=9224
https://news.ycombinator.com/item?id=9224
Classic link. Not the same thing though. I mean try telling a competent agent something like
> Let's build a webapp that allows the user to choose an audio file, we slice it into timestamped word regions with transformers.js + onnx-community/whisper-base_timestamped (with language selection). The user can then scroll through the waveform (that shows the found word regions), synchronized with the transcript, and select word regions in the transcript or in the waveform, and fine-tune the selection if they need to, and then export the slice as WAV.
and see what pops out.
Is it a whisper wrapper? Or does it use custom-trained models?
Vocal Slice uses whisper for the speech-to-text step, then performs its own logic on the transcription and audio, search, take matching, audio slicing, etc.
How can we use this for YouTube?
Thanks for your question - currently Vocal Slice only supports audio files, so you could download your youtube clip, extract the audio and run it through Vocal Slice. I have plans to natively support video in the future.
so it's for selecting cuts that make it into a video for example?
Yeah that's one great use case - you could search for text in a movies audio stream and add that to a video clip for example.
can you make youtube poops with it?
100% it is encouraged!
Cassette Boy would love this.
a great use case example!