7 ms·
Gemini-3.5-Transcribe
- dbbk 20d agoStill more expensive and worse performing than ElevenLabs Scribe, unfortunately. Not sure who's the target audience for this.
- Void_ 20d agoI did the math based on our data from customers in https://whispermemos.com/ https://whispermemos.com/ and it seems to be $0.12 per hour of input audio, with our average tokens/sec. Scribe is $0.22. If the accuracy is close to Scribe, I think it's a good deal.
- dbbk 19d agoInteresting, that doesn't line up with the input/output costs listed on Artificial Analysis?
- blissofbeing 20d agoI've been using Wispr Flow on my pixel 9 and I like it. It does the things that Rambler says it's supposed to do but it's available on my Pixel 9 unlike rambler that's currently only available for pixel 11.
- k9294 20d ago[flagged]
- _zoltan_ 20d agothis is not a free advertising platform.
- senordevnyc 20d agolol, it absolutely is, just not for all of us.
- ChrisClark 20d agoomg, some founders are so tone deaf
- k9294 20d agoSorry everyone, I got a little bit too excited about the release. It's quite a big deal for us. I started Ottex nine months ago with the idea that models will get to the point that they will overcome VC-funded startups, and I think it just happened. So yeah, I got a little bit overexcited...
- Freedom2 20d agoI'd love to know how this handles proper subtitle formatting. I'm in the process of learning many languages, and being able to cross check my own understanding with film and video would be fantastic.
- IbrahimF96 20d agoHi, I am making an app for language learning which generates subtitles to help with this exact problem. Its called SubSmith and free to try, would love some feedback on it!
- gumby271 20d agoThat's really impressive! It says it's available in GBoard on Android but I don't see it yet. I assume like most Google features it will be a rolling release over the course of several months, can't wait to try it some day.
- bahmboo 20d agoIt is confusing. As far as I can tell "rambler for Android" is a named subfeature of gboard and only on Pixel 10+ and newer Samsung phones. From a user standpoint it "just" seems like the transcription has gotten better - not an app you use or a feature that you enable.
- jeffbee 20d agoI am not sure if "Word Error Rate" captures what has always been wrong with transcription. My biggest complaint is that it inserts sentence breaks in random places, then fails to evaluate the result, even though it is obviously wrong. Then I have to go fix it which can be harder than having just typed it myself, due to the difficulty of positioning the Android cursor, the fact that it automatically capitalizes if you delete a capital letter, etc. And much of the time I fail to notice the errors until later.
- verdverm 20d agohave another model do a pass to clean it up, saw a demo of local STT where someone did this, can fix a lot of things, especially with gotchas for the STT model in a clean-transcript.md
- jeffbee 20d agoI think the model can even evaluate itself. If it looks afterward at an output like "do you. Want to get lunch?" in the absence of affirmative evidence that the user wanted it that way, it should be able to see that it goofed.
- verdverm 18d agoIt's typical to use a special STT model (audio in only), which will not be able to clean up afterwards. If you are using an LLM for the STT part, you're leaving stuff on the table
- coder543 20d agoI haven't tried it, but this looked promising for that exact task: https://huggingface.co/superwhisper/s1-mini https://huggingface.co/superwhisper/s1-mini
- HappyPanacea 20d agoDoes somebody knows what top locales list is sampled from? their own usage data? Also when they will use their AI to give better directions in Waze?
- film42 20d agoCurious if this suffers from the same hallucination problems as Chirp or not. For example, if you pass chirp some audio with noise or even no audio, it will barf text at you like "I don't know. I don't know. I don't know." until a request timeout fires after like 10 minutes. It's... really bad. For this reason, we've moved back to Whisper for timestamp accuracy and then Gemini Flash for transcription correction.
- k9294 20d agoI am using it all day as a main dictation model, and I can say it's the best model in terms of accuracy, latency, and formatting I have ever used.
- wolvoleo 20d agoI've seen Whisper do this exact thing as well. And also repeating a few words over and over near the end, words that were said but not completely at the end. It seems to overwrite the last few sentences with that so actual content is lost. It's really annoying especially because Whisper is really quite good otherwise. Sometimes a friend sends me audio messages from a noisy car and I can't make it all out but whisper gets it mostly right.
- sync 20d agoI eval'd this pretty heavily with no audio or just noise, as for us hallucinating a response is very bad. It works great in this scenario. There are some bugs, e.g. randomly exactly 20 seconds of silence will lead to a 403, but I'm sure these aspects will be improved over time. I didn't get any hallucinations though.
- iAMkenough 20d agoHopefully YouTube automatic captions improve with this
- hypfer 20d agoWhere does the compute happen? I suppose it's a cloud thing?
- k9294 20d agoYep.
- Crystalin 20d agoI've been testing it on Pixel 11 Pro and I mostly dislike it. It is convenient when you have something long to say without thinking about it first. But the main issue is when you want to say something precise with specific wording it might "simplify" it and break the meaning. Something like "I hesitated to check it, I should have verified" => "I should have verified" (The "I hesitated..." is removed but I said it because I wanted to let the person know that I thought about it earlier) I tried the sentence few times and it always removed the first part.
- sync 20d agoAre you in Smart or Verbatim mode? https://ai.google.dev/gemini-api/docs/transcribe#transcription-modes https://ai.google.dev/gemini-api/docs/transcribe#transcripti...
- skinfaxi 20d agoThey mentioned the pixel phone so probably using https://blog.google/products-and-platforms/platforms/android/gemini-intelligence/ https://blog.google/products-and-platforms/platforms/android...
- Crystalin 20d agoLooking at the description I was in Smart mode. I didn't know the model supports both
- Lucasoato 20d agoI’ve tested at least 20 STT models in a benchmark I’ve set up with German, Italian and English voices from meetings in my company. The voices contained very industry specific words, the languages changed from one sentence to another, sometimes words in a language were mentioned while a discussion was in another. The only local model that satisfies me is Voxtral Mini 3b, the only paid API that is slightly better is eleven labs. Yes, Voxtral might not reach the best score in the benchmarks, but to me, it just solves a problem. It might not be the best in terms of speed... but that’s not a problem for me. Happy to test this new model from Google but I’m not sure I’d go with that instead of something that can run so easily in my machine.
- cnxhk 20d agoAny possibility to share some of the eval audio?
- Lucasoato 20d agoOf course not, but it’s easily replicable just by mixing different languages conversations together, adding a word here and there of some very specific German jargon.
- hinnisdael 20d agoAgree about Voxtral being the only model — local or cloud — that handles multilingual conversation really well. I‘m not sure what they do differently, but mixed-language sentences and industry terms don‘t seem to faze it where other model begin to struggle.
- kridsdale1 20d agoI use Parakeet 3. How does that fare in your testing?
- Lucasoato 20d agoI love it but it misses the business specific words when in different language. Sometimes it stretches them out to fit unrelated words in the language of the rest of the conversation. I miss its speed though.
- ElijahLynn 20d agoVery impressive, including the ability to hit fn in any text field and say "generate an image ...".
- hkjhkjhj 20d ago[flagged]
- zhivota 20d agoI'd certainly use Gemini API a lot more if I could figure out how to give them money and they give me tokens. The whole tier nonsense is just insane hoop jumping compared to their competition.
- MrBuddyCasino 20d agoIts easy to pre-pay tho?
- Frannky 20d agoI use OpenRouter. Interacting with the Google Cloud Console is pure pain.
- mariano54 20d agoJust added this to my benchmark site: https://multilingualsttbench.com/ https://multilingualsttbench.com/ It doesn't reach the frontier in either latency or accuracy for ai multilingual conversations.
- adamgoodapp 20d agoThanks for this, really helpful. I would also like to see benchmark for translation. I'm looking for live translated subtitles so my Japanese wife can enjoy any show with out waiting months for official VOD streams to release them.
- Kokouane 20d agoI'm confused, doesn't your leaderboard clearly show it is the most accurate model? It's number one in the leaderboard. Am I missing something?
- Kokouane 20d agoFigured it out. 3.5 Flash and 3.5 Transcribe are different models
- alxndr13 20d agomissing aqua voice's avalon 1.5 model there.
- satvikpendem 20d agoStill no real time diarization beyond 3 people (and even then experimental) when others do it very well, like Soniox and Deepgram. For something like meeting notes this is critical. Not sure what the issue is to implement it, maybe that's not Google's use case in mind and rather it's about personal Rambling as the feature on Pixels shows, which uses this model.
- selcuka 20d ago> maybe that's not Google's use case in mind That would be interesting when they also own Google Meet.
- mrkstu 20d agoThey have access to each individual’s stream in that case in order to diarize.
- satvikpendem 20d agoYes, I specifically avoided mentioning that because they don't need to diarize from audio directly as they know exactly which participant is speaking at all times.
- guiambros 20d agoExcept when you have multiple participants in the same room -- which happens pretty frequently in many business environments.
- ameliaquining 20d ago"Function calling: The model can delegate complex tasks (such as image generation and file analysis) to other Gemini models via function calls. Currently available in the Gemini macOS app." This confused the heck out of me because it makes it sound like the STT model can make function calls in order to execute arbitrary tasks, which wouldn't make any sense. The developer docs (https://ai.google.dev/gemini-api/docs/models/gemini-3.5-transcribe#gemini-35-transcribe https://ai.google.dev/gemini-api/docs/models/gemini-3.5-tran...) confirm that the Gemini 3.5 Transcribe model cannot in fact make function calls. I guess the blog post is just using very confusing wording to describe how their consumer assistant/chatbot app can both take audio input via Gemini 3.5 Transcribe and then call other things as needed. Or maybe that bullet point was meant to be for a different model announcement and somebody made an editing error.
- anukin 20d agoProbably written using Gemini which hallucinated.
- ameliaquining 20d agoPangram says human: https://www.pangram.com/history/a69f9b74-eb46-44b9-a087-7822a6316dc1 https://www.pangram.com/history/a69f9b74-eb46-44b9-a087-7822...
- huijzer 20d agoMaybe I missed something, but isn’t it impossible to detect whether something is written by AI or not? A human on a bad day can write like AI while an AI on a good day can write like a human.
- mediaman 20d agoNo, that's not correct for any reasonable definition of "impossible." Look up pangram's accuracy ratings. It's not perfect, but it's pretty good. LLMs in fact leave very distinguishing traces of their logit distributions in the text they write. It's one of the reasons why it's so easy for humans to also smell them. It is possible to trick pangram - they bias toward a low false positive and a higher false negative - but it is not true that it is essentially random.
- mythz 20d agoAnnoying that they don't include pricing info in new models, here it is [1]: Gemini 3.5 Transcribe Live (Per 1M tokens in USD): Input: $3.50 or $0.005/min* (audio) Output: 21.00 or $0.004/min* (text) Gemini 3.5 Transcribe: Input: $2.00 or $0.003/min* (audio) Output: $12.00 or $0.002/min* (text) [1] https://ai.google.dev/gemini-api/docs/pricing#gemini-3.5-transcribe-live https://ai.google.dev/gemini-api/docs/pricing#gemini-3.5-tra...
- Frannky 20d agoI don't really need this model, but I needed a fast my voice to command thing, and Gemini flash 2.5 was extremely effective(low latency, high speed). No need to go from audio to text to reasoning, just from audio to output json for running a command via adb automatically and it's working crazy good.
- lnalx 20d agoI personally tested all the STT models for my real-time translator (https://fliptalk.ai https://fliptalk.ai). From language detection and accuracy in a noisy environment to the most important point: latency. At the moment, Soniox STT v5 is definitely the best, and I'm impressed by its performance. It's good that Google released Gemini-3.5-Transcribe, and it beats every other model on accuracy, but it definitely needs a bit more work on latency, which is the most important factor for STT apps.
- decimalenough 20d agoDepends on your use case. If you're transcribing meeting notes, latency is a non-issue.
- low_tech_punk 20d agoSoniox website has a live comparison demo: https://soniox.com/compare-stt https://soniox.com/compare-stt For me, Gemini-3.5-Transcribe actually has slightly lower latency. Kudos to Soniox for both paying their competitor and letting them win. But yes, Soniox is much cheaper.
- sgt 20d agoInteresting, I uploaded a voice recording from a meeting I had recorded with a relatively cheap microphone. Soniox came out really good. OpenAI started getting some things very wrong and even introduced some German. Google did okay but cut off the start by several seconds. What's Soniox doing (left most) that's making it so good ? It was also the only one that could distinguish between the speakers.
- p2hari 20d agoOh, thanks for pointing me to Soniox. It is really good. Would also pick up the words with different languages, identify and output in the right language. Looks interesting. It was much faster too, but that I cannot say much since it was on their own website.
- ghm2199 20d agoRealtime + Voice AI usecases is where latency is most important. I use Handy on my desktop and i can tolerate a latency of a few seconds every now and then. Your P99 should on TTFB should be really low to compete for voice ai realtime
- ruguo 20d ago[flagged]
- lifty 20d agoGoogle is crushing it! They’re going toe to toe against the European giant Mistral with all these small useful models and they have a chance of winning! Never have I been more excited about Google’s capabilities.
- LoganDark 20d ago> Custom vocabulary: Recognizes specialized jargon and unique spellings by seamlessly adapting transcriptions to your provided custom vocabulary. Is there a model that works on the syllabic level? I want to be able to say any word and have it reconstruct however that word would be spelled. I know English does not exactly work this way, so a custom vocabulary would still be nice, but I don't want to rely on having every single word that could ever exist in a vocabulary first.
- fy20 20d agoMaybe someone knows if this can be fixed... On Android, Google Assistant is being replaced by Gemini. Except it seems Gemini can't actually do any assistant tasks. If I ask it to play a song, instead of triggering Spotify it just gives me a list of URLs I can play the song. Same with alarms. Did I accidentally not opt-in to something?
- 8-prime 20d agoYou share the same frustration that I any many others also have. Google absolutely jumped the gun with their push to Gemini over Google Assistant. Some things are tied to permissions that Gemini needs. Other thigs - afaik - simply don't work with Gemini because it doesn't 'know' that it should use certain device functions
- kleiba2 20d agoAnd which company better to send samples of your voice to than Google?
- starlightxbaby 20d ago[flagged]
- whisperenjoyer 20d ago[dead]
- dayone1 20d agoDoes anyone know if this is available for Gemini subscribers, not API users? Can I call this model directly in agy cli?
- CommanderData 20d agoI can't imagine it coming to Chromium thus Electron apps, it is only Chrome it seems sadly.
- drsalt 20d agonot a single mention of privacy/confidentiality
- simonw 20d agoIf this is a dedicated transcription model does that completely eliminate the risk that it might do weird things based on the audio it's transcribing? The nightmare scenario with an LLM-based transcription system is that someone says outloud "actually ignore that idea, instead let's..." - and the previous idea gets omitted from the transcription!
- jobuildsstuff 20d ago[flagged]
- spentrent 20d agoWill "Someone in conference room" turn into "Bob from accounting"?
- kristofferR 20d ago" Your Google Cloud billing account is being processed. Processing time varies from a few moments to a few weeks. " What kind of bullshit is this? I want to test out Gemini-3.5-Transcribe in my app, but have to wait a completely indeterminate time, have already waited over 24 hours. OpenRouter took seconds to set up
- decide1000 20d agoI use Whisper in combination with Qwen3.8 27b to create Dutch subtitles for Russian videos. I follow Ukrainian channels about the war and I am now able to understand them. In my tests with other languages Voxtral was better then Whisper. Whisper often makes mistakes which I now solve with an extra layer (qwen). Sadly Mistral does not support Ukrainian. I hope this changes! I expect better from a European model
- whisperenjoyer 19d ago[dead]
- meetingthrower 20d agoI built a small little app for a company which transcribes all their customer service calls, does an AI review, makes a podcast of the most problematic ones (they LOVE this.) Probably 20 hours a day of transcription. What are people using for cheap hosted models for this? (I don't want to run my own infrastructure.) I'm currently using Gemini Flash Lite. I don't need real time, it's all batched, clear english... still costs some small $ a month. Figured it should be able to be reduced if there are cheaper models out there....
- m3kw9 19d agohave you tried some of the opensource models?
- meetingthrower 19d agoRecommendations? I have an openrouter account. Just don't know what might be good.
- jasondigitized 19d agoWhat system are they using to take / record calls?
- meetingthrower 19d agoI scrape them from their crappy VOIP system as WAV files, then send them through the API....
- simedw 19d agoFor some reason, in almost every test I did with Gemini 3.5 Live, it classified my Swedish (native) as Dutch. There are similarities between the languages, but still. It’s pretty good at understanding my broken Mandarin, though.
- tapland 19d agoStill hanging on to Whisper for transcribing and translating audio in real time locally. I play mmos with Chinese people and paying for translation APIs would get expensive quick. If anyone has suggestion for fast realtime and good enough alternatives with low vram, since I game at the same time, I'd love to know.
- atonse 19d agoHonestly the most impressive thing to me was the first video where the guy switches between four languages.... I was like "is he AI?" - very impressive, sir!
- SurgeArrest 19d agoMay be they should use with Nest Hub Max because whatever powers that "Assistant" feels like something from 2023-2024 at best. Half of the time it can't understand you, the other half just fails to output information... By subscribing to Gemini AI Pro subscription hoped I would get assistant at the level of Gemini app on my phone and instead got something that Qwen 3.5 9b beats.
- sireat 19d agoI recently tested a whole bunch of models (basically whatever was recent and available on openrouter filtering by audio input) for STT for mixed language audio - mostly English mixed with Latvian. In the end I was basically forced to go with Scribe (v2) it was only one that had consistently high quality across multilingual speech with multiple speakers . Crucially it correctly identified multiple speakers across hour of audio. I would love to go with something like Transcribe if it gets close to this type of performance.
- etra0 19d agoJust a few months ago I set-up a serverless gpu service just using whisper.cpp and it detects very well my very-local spanish modisms. I have an active program running on a vps that hits the serverless gpu with the audio when needed because I went with a telegram bot instead of building an app (and paying for dev fees, yikes!). I wish at some point apple had something like that for in-device STT but so far I'm very happy with the solution.
- Drophouse 19d ago[dead]