Rendered at 14:55:29 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
Shershah1024 2 hours ago [-]
Hi all. I am Shahir. This was my experiment. First of all - Thanks to whoever posted this on HN. It came as a surprise.
Just to clarify - This started as just an experiment to see if we can use the fact that text and audio share a vector space in Embedding Gemma 2 to convert audio to text. Had to hack around a few things, but, it worked. Speed was just a side benefit. Was not aiming for it.
The accuracy is good, but not in the range of cloud models, but decent enough for everyday use, and this is just a couple of days work. The WER can improve a lot with more training on varied data. Will try that next.
I will opensource the whole thing soon. Might need a few days though. Also, releasing the TTS version for this soon.
Happy to answer any questions you have.
xnx 52 minutes ago [-]
Cool experiment. It will be amazing if this can match the WER of the best open weight models at better speed.
People aren't paying enough attention to how great EmbeddingGemma 2 and Gemma 4 are.
usern20260720 19 minutes ago [-]
can this tool transcribe latin language texts? McWhisper allegedly does it but it outputs gibberish
cabarltos 58 minutes ago [-]
Thanks for putting this out there!
prodigycorp 1 hours ago [-]
can you turn this into a decision model
drakythe 1 hours ago [-]
What would a decision model add to and audio transcriber?
cheschire 25 minutes ago [-]
Reducing cost for a large scale corporation call center. Press or say one.
Basic ML is probably even cheaper. But LLMs might have higher accuracy. “Uhh one I think” = 1
whstl 25 minutes ago [-]
For people interested in super quick models:
"One cool trick" I like for running those kinds of models very quickly on macOS is converting them to CoreML so they run in the Neural Engine.
I had amazing success with Spleeter, Demucs, MVSEP-MDX23-music-separation, DeepFilterNet, Whisper, Hunyuan3D for 3D mesh generation. Plus a lot of others for both work and leisure.
I also convert the code to Swift without dependencies, but that's a personal preference.
The runtime on some of them went sometimes from 60s to 10s. And since it's just Swift, I can just give them to non-technical co-workers and they don't have to setup a whole environment.
If you don't wanna do this by hand, this is something that existing LLM models are good at. Even Luna can do it without much fuss.
sbdaman 53 minutes ago [-]
Maybe it's just a COVID time blending together effect but it still blows my mind how quickly we went from "automatic audio transcriptions are not even worth having on" to "an hour of audio transcribed in 13s locally."
Pre-pandemic it seemed like even the very best in class auto-transcription features were just unusably bad. I guess that was like 7 years years ago though.
gambiting 31 minutes ago [-]
I'm just waiting for someone to tell the YouTube team. Hopefully 10 more years and someone over there will realize it's a thing.
kingleopold 23 minutes ago [-]
It's funny cause someome needs to remind them that their company is owned by google + somebody needs to care about that part. It might even take longer than 10 years.
f6v 50 minutes ago [-]
[dead]
saaaaaam 1 hours ago [-]
I recently swapped to using Parakeet v3 on my old M1 Macbook Air, using MacWhisper. It was so fast I thought it had failed. As in, by the time I clicked "transcribe" and looked at my screen again a few seconds later the transcript had finished. Absolutely insanely fast.
llimllib 3 hours ago [-]
Apple's on-device transcription model is about 3x slower than their reported results on my more powerful computer, with a similar error rate in my test (which I consider pretty good results)
update: doesn't seem like they provide their model and I don't feel like trying to dig it out of their app bundle, so I'm not going to test this one
s3p 2 hours ago [-]
Is it even worth using apple's? Maybe they have improved it but when they first started using it to transcribe voice memos it was laughably bad (in my experience). I defaulted to whisper.cpp
ChrisMarshallNY 31 minutes ago [-]
I suspect Apple's strengths are in multi-language support.
I use their on-device translation in my apps. Works pretty well. ChatGPT's context-aware translation is better, but it's not on-device.
llimllib 1 hours ago [-]
apple's is 5x faster than `large-v3-turbo` with whisper.cpp and in my test had somewhat worse performance, so it's a tradeoff
bigtex 2 hours ago [-]
With MacWhisper you can select the Apple models, it took almost 4x as long to transcribe the same file versus this software and MacWhisper/Parakeet3.
llimllib 1 hours ago [-]
interesting! I only tested against Parakeet v2 because my agent suggested it would be faster for English transcription (v3 is multilingual but somewhat slower, it suggested)
On my bench, parakeet-tdt-0.6b-v2 is 2x slower
I wonder if you're using different Apple models when you go through that app? I don't know
bigtex 2 hours ago [-]
I tested a 33 minute local audio only file on a M4 Mac Mini with 24GB of Ram and it took 6.75 seconds to transcribe it using this software. Using MacWhisper/Parakeet3 it was about 7.5 secs using a stopwatch.
fdb 3 hours ago [-]
Is it open source? They mention something about the Apache License but I can't seem to find a link to GitHub or something.
cstever 3 hours ago [-]
I only see that EmbeddingGemma 2 is Apache 2.
"EmbeddingGemma 2 is made by Google DeepMind and released under Apache 2.0. This is an independent project, not affiliated with or endorsed by Google."
ghboo0927 2 hours ago [-]
The reference to Apache 2.0 on the page refers to Google's base model, not this app. Based on the page alone, the app is free, but the source code is not open-source.
Yes. Planning to train a multilingual reader. Have not done a prototype yet.Will do one once I have the right dataset
samuelbalogh 1 hours ago [-]
Awesome work, thank you!
2 hours ago [-]
alberth 3 hours ago [-]
As someone who knows nothing on how this works, does this mean they have to pay the audio at 200x playback speed to transcribe this fast.
Or is there another approach?
treyd 3 hours ago [-]
It's not being played through a desktop audio stack, they're operating on the waveform decoded from the audio file. There's no constraints relative to the "natural speed" of the source, just like for video transcoding, it's limited by compute.
radlad 2 hours ago [-]
Yes but... Frame rate matters with video, which is arguably the video's speed.
sorou 3 hours ago [-]
during transcription, nothing is actually "playing" the file in the way a human would to listen to it. the audio is stored as a sequence of bytes representing the amplitude of the physical sound waves over time (the recording device measures, or "samples", the amount that its microphone is vibrating and records that amount, some tens of thousands of times per second) with some metadata and compression applied so that other software knows how to interpret the bytes correctly. this can be given to an audio program to convert the bytes back to physical sound waves using your speakers, but it's just bytes at the end of the day, so any software that can read bytes out of the file can do so at whatever speed it wants, regardless of how long the actual recording is.
so you have a file containing bytes representing the sound wave, and you have existing recordings (read: sound waves represented as sequences of bytes) of people speaking in different languages and contests, and you can feed all those existing recordings into a model along with corresponding transcriptions, and you can get a model out which can identify patterns in recordings of people speaking to then go backwards and see a pattern of sound waves and correctly associate it with a specific token, no need to play anything back. pauses, stutters, etc. are themselves patterns of sound waves that can be identified and worked around to identify just the meaningful parts of speech.
in principle, you could probably learn to do this too, by sitting down for hundreds of hours meticulously starting at waveforms, and this is, very roughly, what signal analysts do: look at waveforms and try to identify patterns. ai just leverages this principle to turn mundane speech into text.
cheschire 3 hours ago [-]
Probably significant parallelism. Split the audio into a lot of slightly overlapping portions to ensure the transitions between words are caught. I’m sure there’s some relatively simple ways to ensure you don’t double up on words.
drob518 3 hours ago [-]
Yea, I was wondering how they dealt with word boundaries and different word audio lengths generally (short and long). Your point about parallelism is interesting. Doing what they say they are doing, this should be ridiculously parallel, possibly with a reduce step at the end to remove redundant words if their slicing algorithm allows for that to happen. But yea, extremely parallel.
satvikpendem 2 hours ago [-]
No more than an AI media generator creating the media in realtime.
yaanfpv 2 hours ago [-]
I think this seems to target English only I believe. Pretty good but thinking os using some of these techniques on our multi-lingual ASR models !
Shershah1024 1 hours ago [-]
Only English for now. I dont see any roadblocks for at least the major European Languages(Not sure about German though with their compound words)
lukko 3 hours ago [-]
Interesting... is it overkill to use a multimodal model like this on just audio? Could the model size be decreased dramatically, or does the multimodality help the accuracy and speed in some way?
pjc50 23 minutes ago [-]
The benefit is that you can take input in one "mode" and extract output in another; that is, the input is audio mode and the output is text mode. The intermediate representation is the "embedding".
sd9 2 hours ago [-]
Multimodality seems to just help with everything
coldtea 3 hours ago [-]
Correctly transcribing requires context too, no?
ranger_danger 2 hours ago [-]
I suppose it depends on the words in question... language knowledge and general world knowledge may not always both be necessary to transcribe accurately.
amelius 2 hours ago [-]
Sounds great if you could use it as a server. Otherwise, why would you want this?
etienne_l 2 hours ago [-]
[dead]
martianvoid 3 hours ago [-]
Pretty cool application of an embedding model
chronogram 3 hours ago [-]
What about the quality of the transcription?
walrus01 3 hours ago [-]
They say its word error rate is about 5%, so not great. Interesting as a proof of concept. The model this is based on is REALLY tiny.
koolba 2 hours ago [-]
Too bad the errors for things like this are not truly independent.
Otherwise could run two passes to get down to .25% or three passes to .0125%. Or four passes to .000625%!
That’d be .02 seconds of error per hour!
robotswantdata 3 hours ago [-]
WER is not good enough for professional use, fine for hobby projects
vladimirralev 3 hours ago [-]
Any info on the param size and quantization of each part and in total?
mikemzz1 3 hours ago [-]
interesting approach. But training ctc head on top of gemma's audio encoder will most likely produce better results and faster inference
Shershah1024 1 hours ago [-]
That's close to what this is: the reader is a CTC head on the audio encoder, but its output words are EmbeddingGemma's own text embeddings for ~229k words, instead of a separately learned vocabulary. I also tried a plain letter-level CTC head on the encoder. It did a bit better on clean audiobooks (7.0 vs 8.2 WER at the time) but much worse on harder audio like a lecture (16.5 vs 10.8), mostly from misspellings.
spiderfarmer 2 hours ago [-]
So Marco Arment can sell some more of his Mac Minis.
Aboutplants 2 hours ago [-]
First thing I thought of lol
jdw64 3 hours ago [-]
It would be nice if Korean also worked. I’d like to experiment with a Korean model too.
Imustaskforhelp 3 hours ago [-]
This has come at just the right time.
Does this work for Hindi or other non English languages though?
I wanted to recently transcribe Hindi + English (mixed) Audio and I used this model[0] and created a custom transcribe command which can give text over input of mp4 files
This took me roughly around an hour or more or around this time frame for a 90 minute transcription source. I would consider myself to be slightly bored during this phase.
I would really appreciate it if something like the project could be made native for a lot more languages but I think that the approach of embedding models and using them in such a way is pretty crazy. Kudos for it and I will try it
out!
I can think of many such use cases where Speech to text of even large audio files becomes near instant. As I think someone on hackernews once quoted whom I am very paraphrasing
An order of one or two magnitude of an already existing thing stops becoming a quantity upgrade but rather becomes a quality upgrade in it of itself as it provides access to workflows which previously wouldn't have been feasible.
Yeah I will have to try whisper as well. I didn't try it because I wanted to tinker more and also I had thought that whisper was an older model but I think that if it gets the job, then it gets the job done and I will be trying it out on the text that I mentioned, thanks for mentioning it as I will give it another go :-D
tinykerneldev 2 hours ago [-]
TrueScribe supports tranlating Hindi transcripts in both directions. Not free though.
arbor-group 50 minutes ago [-]
[flagged]
octowatch 3 hours ago [-]
[flagged]
mengyuxuan 3 hours ago [-]
[flagged]
bawana 2 hours ago [-]
I wonder if this approach can be used to transcribe the soounds made by dolphins and whales
Just to clarify - This started as just an experiment to see if we can use the fact that text and audio share a vector space in Embedding Gemma 2 to convert audio to text. Had to hack around a few things, but, it worked. Speed was just a side benefit. Was not aiming for it.
The accuracy is good, but not in the range of cloud models, but decent enough for everyday use, and this is just a couple of days work. The WER can improve a lot with more training on varied data. Will try that next.
I will opensource the whole thing soon. Might need a few days though. Also, releasing the TTS version for this soon.
Happy to answer any questions you have.
People aren't paying enough attention to how great EmbeddingGemma 2 and Gemma 4 are.
Basic ML is probably even cheaper. But LLMs might have higher accuracy. “Uhh one I think” = 1
"One cool trick" I like for running those kinds of models very quickly on macOS is converting them to CoreML so they run in the Neural Engine.
I had amazing success with Spleeter, Demucs, MVSEP-MDX23-music-separation, DeepFilterNet, Whisper, Hunyuan3D for 3D mesh generation. Plus a lot of others for both work and leisure.
I also convert the code to Swift without dependencies, but that's a personal preference.
The runtime on some of them went sometimes from 60s to 10s. And since it's just Swift, I can just give them to non-technical co-workers and they don't have to setup a whole environment.
If you don't wanna do this by hand, this is something that existing LLM models are good at. Even Luna can do it without much fuss.
Pre-pandemic it seemed like even the very best in class auto-transcription features were just unusably bad. I guess that was like 7 years years ago though.
I'll test theirs too if I get around to it
Here's a sample CLI wrapping the apple model and providing a simple benchmark: https://github.com/llimllib/transcribe
update: doesn't seem like they provide their model and I don't feel like trying to dig it out of their app bundle, so I'm not going to test this one
I use their on-device translation in my apps. Works pretty well. ChatGPT's context-aware translation is better, but it's not on-device.
On my bench, parakeet-tdt-0.6b-v2 is 2x slower
I wonder if you're using different Apple models when you go through that app? I don't know
"EmbeddingGemma 2 is made by Google DeepMind and released under Apache 2.0. This is an independent project, not affiliated with or endorsed by Google."
Or is there another approach?
so you have a file containing bytes representing the sound wave, and you have existing recordings (read: sound waves represented as sequences of bytes) of people speaking in different languages and contests, and you can feed all those existing recordings into a model along with corresponding transcriptions, and you can get a model out which can identify patterns in recordings of people speaking to then go backwards and see a pattern of sound waves and correctly associate it with a specific token, no need to play anything back. pauses, stutters, etc. are themselves patterns of sound waves that can be identified and worked around to identify just the meaningful parts of speech.
in principle, you could probably learn to do this too, by sitting down for hundreds of hours meticulously starting at waveforms, and this is, very roughly, what signal analysts do: look at waveforms and try to identify patterns. ai just leverages this principle to turn mundane speech into text.
Otherwise could run two passes to get down to .25% or three passes to .0125%. Or four passes to .000625%!
That’d be .02 seconds of error per hour!
Does this work for Hindi or other non English languages though?
I wanted to recently transcribe Hindi + English (mixed) Audio and I used this model[0] and created a custom transcribe command which can give text over input of mp4 files
This took me roughly around an hour or more or around this time frame for a 90 minute transcription source. I would consider myself to be slightly bored during this phase.
I would really appreciate it if something like the project could be made native for a lot more languages but I think that the approach of embedding models and using them in such a way is pretty crazy. Kudos for it and I will try it out!
I can think of many such use cases where Speech to text of even large audio files becomes near instant. As I think someone on hackernews once quoted whom I am very paraphrasing
An order of one or two magnitude of an already existing thing stops becoming a quantity upgrade but rather becomes a quality upgrade in it of itself as it provides access to workflows which previously wouldn't have been feasible.
[0]: https://huggingface.co/parismitaglobalsolutions/indicconform...
https://openai.com/index/whisper/
https://github.com/openai/whisper