51–60 of 274 posts
Earlier quoted context omitted.
I'm the simplest way - if you allow AI to make decisions, you're responsible. Like this https://bc.ctvnews.ca/air-canada-s-chatbot-gave-a-b-c-man-th... So far we're doing pretty good with that idea globally (I've not seen any case going the other way in court)
I mean how would it work, if you tried to hold the AI liable?
Just the other day there were news about how many mistakes the AI-assisted journaling caused – wrong names, wrong diagnoses, typos (buksmärta -> kuksmärta which is hilarious but serious). Some of it is surely teething problems, but unless there is a robust check upon implementation it might just add another layer of inefficient new public management make-work to the system. https://sverigesradio.se/artikel/ai-journal…
> typos (buksmärta -> kuksmärta which is hilarious but serious) To save people looking it up, that one-char difference changes "abdominal pain" into "cock pain". Wow.
That is the main problem with the AI: it is close enough, but never there.
Earlier quoted context omitted.
I wouldn't be surprised if Region Blekinge were using something much worse and much more expensive than Whisper for their transcription. I've been transcribing A LOT of SR (Swedish Radio) shows as part of https://nyheter.sh/ , and Whisper (self-hosted) has been very accurate.
Tangent here: really? I've found base Whisper has concerning error rates for non-US English accents; I imagine the same is true for other languages with a large regional mode to the source dataset. Whisper + an LLM can recover some of the gaps by filling in contextually plausible bits, but then it's not a transcript and may contain hallucinations. There are alternatives that share Whisper internal states with an LLM…
I can't speak to how it performs outside of production quality audio, but in the hundreds of hours of subtitles that I've generated I don't think I've seen a single error.
Earlier quoted context omitted.
I wouldn't be surprised if Region Blekinge were using something much worse and much more expensive than Whisper for their transcription. I've been transcribing A LOT of SR (Swedish Radio) shows as part of https://nyheter.sh/ , and Whisper (self-hosted) has been very accurate.
Tangent here: really? I've found base Whisper has concerning error rates for non-US English accents; I imagine the same is true for other languages with a large regional mode to the source dataset. Whisper + an LLM can recover some of the gaps by filling in contextually plausible bits, but then it's not a transcript and may contain hallucinations. There are alternatives that share Whisper internal states with an LLM…
Traditional ASR systems struggle when English (or any language) is spoken with a heavy accent, often confusing it with another language. Whisper is also affected by this issue, as you noted.
The root of this problem lies in how language detection typically works. It relies on analyzing audio via MFCC (Mel Frequency Cepstrum Coefficient), a method inspired by human auditory perception.
MFCC is a part of the "psychoacoustic" field, focusing on how we perceive sound. It emphasizes lower frequencies and uses techniques like normalized Fourier decomposition to convert audio into a frequency spectrum.
However, this approach has a limitation: it's based purely on acoustics. So, if you speak English with a strong accent, the system may not understand the content but instead judge based on your prosody (rhythm, stress, intonation).
With the team at Gladia, we've developed a hybrid approach that combines psycho-acoustic features with content understanding for dynamic language detection.
In simple terms, our system doesn't just listen to how you speak but also understands what you're saying. This dual approach allows for efficient code-switching and doesn't let strong accents fall through the cracks. The system is based on optimized Whisper, among other models.
In the end, we managed to solve 99% of edge cases involving strong accents, despite the initial Whisper bias there. We've also worked a lot on hallucinations as a separate problem, which resulted in our proprietary model called Whisper-Zero.
If you want to give it a try, there's a free tier available. I'm happy to bounce around ideas on this topic any time; it's super fascinating to me.
Earlier quoted context omitted.
It feels to me like an autopilot problem in the making. "This thing means that you don't have to keep your eyes on the road - but please ensure you keep your eyes on the road, in case of errors"
I found it interesting that your mind went to Tesla's autopilot. My mind went to operating airplanes. Most newer small planes have some form of GPS but you're technically not supposed to use instrument navigation until your certified to do so. I haven't met a single pilot that didn't do so, though. Anyway, it creates the very problem you mentioned but just replace "road" with "outside the cockpit".
What do you mean by this? Not having an IFR rating does not mean you're not allowed to use the navigation aids or the plane's autopilot.
I talked to a therapist who hated doing notes. So I proposed a solution: Use Microsoft's "seeing AI" to describe the children playing during "play therapy". Then have a camera take a picture every few minutes, and then have chatgpt come up with the story based on the pictures and transcript of the audio, transcribed with openai whisper. Business in a nutshell!
Earlier quoted context omitted.
Tangent here: really? I've found base Whisper has concerning error rates for non-US English accents; I imagine the same is true for other languages with a large regional mode to the source dataset. Whisper + an LLM can recover some of the gaps by filling in contextually plausible bits, but then it's not a transcript and may contain hallucinations. There are alternatives that share Whisper internal states with an LLM…
Language detection in the presence of strong accents is, in my opinion, one of the most under-discussed biases in AI. Traditional ASR systems struggle when English (or any language) is spoken with a heavy accent, often confusing it with another language. Whisper is also affected by this issue, as you noted. The root of this problem lies in how language detection typically works. It relies on analyzing audio via MFCC…
Humans also have difficulty with heavy accents, no?
I talked to a therapist who hated doing notes. So I proposed a solution: Use Microsoft's "seeing AI" to describe the children playing during "play therapy". Then have a camera take a picture every few minutes, and then have chatgpt come up with the story based on the pictures and transcript of the audio, transcribed with openai whisper. Business in a nutshell!
Ah, the freedom of having no consideration for privacy or ethics. No one should record children and feed the imagery to a ominous third-party.
I talked to a therapist who hated doing notes. So I proposed a solution: Use Microsoft's "seeing AI" to describe the children playing during "play therapy". Then have a camera take a picture every few minutes, and then have chatgpt come up with the story based on the pictures and transcript of the audio, transcribed with openai whisper. Business in a nutshell!
This is not really how therapists work. It’s how it looks like they work. This is a really big problem with the AI industry. People don’t know the domain well enough and assume it is a fit. Oh look, just throw this at it and done. Simples! Then you find that therapists, of the non quack variety at least, spend years working on how to remove bias from their assessment and learning subtle cues and indicators from their…
How about improving the processes and reduce the amount of forms and admin work instead?