4 ms·

The audio transcript exercise here is particularly interesting from a journalism perspective. Summarizing a 3.5 hour council meeting is something of a holy gra

• WesleyLivesay 1y ago
  I think it appears to have done a good job of summarizing the points that it summarize, at least judging from my quick watch of a few sections and from the YT Transcript (which seems quite accurate). Almost makes me wonder if it is behind the scenes doing something similar to: rough transcript -> Summaries -> transcript with timecodes (runs out of context) -> throws timestamps that it has on summaries. I would be very curious to see if it does better on something like an hour long chunk of audio, to see if it is just some sort of context issue. Or if this same audio was fed to it in say 45 minute chunks to see if the timestamps fix themselves.
• byt3bl33d3r 1y ago
  I’ve been meaning to create & publish a structured extraction benchmark for a while. Using LLMs to extract info/entities/connections from large amounts of unstructured data is also a huge boon to AI-assisted reporting and has also a number of cybersecurity applications. Gemini 2.5 was pretty good but so far I have yet to see an LLM that can reliably , accurately and consistently do this
  • simonw 1y ago
    This would be extremely useful. I think this is one of the most commercially valuable uses of these kinds of models, having more solid independent benchmarks would be great.
• mistercheph 1y ago
  For this use case I think best bet is still a toolchain with a transcription model like whisper fed into an LLM to summarize
  • simonw 1y ago
    Yeah I agree. I ran Whisper (via MacWhisper) on the same video and got back accurate timestamps. The big benefit of Gemini for this is that it appears to do a great job of speaker recognition, plus it can identify when people interrupt each other or raise their voices. The best solution would likely include a mixture of both - Gemini for the speaker identification and tone-of-voice stuff, Whisper or NVIDIA Parakeet or similar for the transcription with timestamps.
• rahimnathwani 1y ago
  For this use case, why not use Whisper to transcribe the audio, and then an LLM to do a second step (summarization or answering questions or whatever)? If you need diarization, you can use something like https://github.com/m-bain/whisperX https://github.com/m-bain/whisperX
  • pants2 1y ago
    Whisper simply isn't very good compared to LLM audio transcription like gpt-4o-transcribe. If Gemini 3 is even better it's a game-changer.
  • crazysim 1y ago
    Since Gemini seems to be sucking at timestamps, perhaps Whisper can be used to help ground that as an additional input alongside the audio.
• ks2048 1y ago
  Does anyone benchmark these models for text-to-speech using traditional word-error-rates? It seems audio-input Gemini is a lot cheaper than Google Speech-to-text.
  • simonw 1y ago
    Here's one: https://voicewriter.io/speech-recognition-leaderboard https://voicewriter.io/speech-recognition-leaderboard "Real-World Speech-to-text API Leaderboard" - it includes scores for Gemini 2.5 Pro and Flash.
• Workaccount2 1y ago
  My assumption is that Gemini has no insight into the time stamps, and instead is ballparking it based on how much context has been analyzed up to that point. I wonder if you put the audio into a video that is nothing but a black screen with a timer running, it would be able to correctly timestamp.
  • simonw 1y ago
    The Gemini documentation specifically mentions timestamp awareness here: https://ai.google.dev/gemini-api/docs/audio https://ai.google.dev/gemini-api/docs/audio
  • minimaxir 1y ago
    Per the docs, Gemini represents each second of audio as 32 tokens. Since it's a consistent amount, as long as the model is trained to understand the relation between timestamps and the number of tokens (which per Simon's link it does), it should be able to infer the correct amount of seconds.
• potatolicious 1y ago
  You really want to break a task like this down to constituent parts - especially because in this case the "end to end" way of doing it (i.e., raw audio to summary) doesn't actually get you anything. IMO the right way to do this is to feed the audio into a transcription model, specifically one that supports diarization (separation of multiple speakers). This will give you a high quality raw transcript that is pretty much exactly what was actually said. It would be rough in places (i.e., Speaker 1, Speaker 2, etc. rather than actual speaker names) Then you want to post-process with a LLM to re-annotate the transcript and clean it up (e.g., replace "Speaker 1" with "Mayor Bob"), and query against it. I see another post here complaining that direct-to-LLM beats a transcription model like Whisper - I would challenge that. Any modern ASR model will do a very, very good job with 95%+ accuracy.
  • simonw 1y ago
    Which diarization models would you recommend, especially for running on macOS? (Update: I just updated MacWhisper and it can now run Parakeet which appears to have decent diarization built in, screenshot here: https://static.simonwillison.net/static/2025/macwhisper-parakeet.jpg https://static.simonwillison.net/static/2025/macwhisper-para... )
    • atonse 1y ago
      I use parakeet daily (with MacWhisper) to transcribe my meetings. It works really well, even with the speaker segmentation.
  • darkwater 1y ago
    Why can't Gemini, the product, do that by itself? Isn't the point of all this AI hype to easily automate things with low effort?
    • vlovich123 1y ago
      Multimodal models are only now starting to come into the space and even then I don’t know they really support diarization yet (and often multimodal is thinking+speech/images, not sure about audio).
• stavros 1y ago
  Just checking, but did you verify that the converted ffmpeg audio wasn't around an hour long? Maybe ffmpeg sped it up when converting?
  • simonw 1y ago
    Here is the file I used, it is the expected length https://static.simonwillison.net/static/2025/HMB-nov-4-2025.m4a https://static.simonwillison.net/static/2025/HMB-nov-4-2025....
• luke-stanley 1y ago
  Simon, sorry I didn't get around to answering your question on post-t5 encoder-decoders from the Markdown Lethal Trifecta prompt injection post. (https://news.ycombinator.com/item?id=45724941 https://news.ycombinator.com/item?id=45724941) Since the plain decoder models stole the show, Google DeepMind demonstrated a way to adapt LLMs,adding a T5 encoder to an existing normal Gemma model to get the benefits of the more grounded text-to-text tasks WITHOUT instruction tuning (and the increased risk of prompt injection). They also have a few different kinds they shared on HuggingFace. I didn't get around to fine-tuning the weights of one for summarisation yet but it could well be a good way for more reliable summarisation. I did try out some models for inference though and made a Gist here, which is useful since I found the HF default code example a bit broken: https://gist.github.com/lukestanley/ee89758ea315b68fd66ba52cd2123c35 https://gist.github.com/lukestanley/ee89758ea315b68fd66ba52c... Google's minisite: https://deepmind.google/models/gemma/t5gemma/ https://deepmind.google/models/gemma/t5gemma/ Paper: https://arxiv.org/abs/2504.06225 https://arxiv.org/abs/2504.06225 Here is one such model that didn't hallucinate and actually did summarise on HF: https://huggingface.co/google/t5gemma-l-l-prefixlm https://huggingface.co/google/t5gemma-l-l-prefixlm
• luke-stanley 1y ago
  I'd lower the temperature and try DSPy Refine loop on it (or similar). Using audio encoder-decoder models and segmentation are good things to try too. A length mismatch would be bad. DSPy has optimisers. It could probably optimise well with length match heuristic, there is probably a good Shannon entropy rule.
• barapa 1y ago
  We have done a lot of work solving this exact challenge at legaide.ai. I was both sad and relieved when gemini 3 didn't deprecate our startup.
• anilgulecha 1y ago
  We've done some transcription exercises. The way to get the timestamps to line is : 1) break up the audio into minutes, and pass it in, one after another, so the chat completions prompt looks like: Here's minute 1 of audio [ffmpeg 1st minute cut out.wav] Here's minute 2 of audio [ffmpeg 2nd minute cut out.wav] Here's minute 3 of audio [ffmpeg 3rd minute cut out.wav] and so on.. The cutting step is simple, and token count is pretty much the same, but the crucial additional detail allows for excellent transcription fidelity time wise. We've also experimented passing in regular TTS (non-llm) transcript for reference, which again helps the LLM do better.
• theshrike79 1y ago
  This is a Google notebook on multimodal transcription from a training session they do: https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/gemini/use-cases/video-analysis/multimodal_video_transcription.ipynb https://colab.research.google.com/github/GoogleCloudPlatform... It goes through the video analysis process and terminology step by step. You need to say "timecode" instead of "timestamp" so the LLM aligns better to AV-stuff more than programming (weird, right?) Basically you need two to three passes to get a proper transcription 1. just listen and identify voices, give them some kind of ID 2. Maybe watch the video and see if the people have name tags on the table in front of them or there's an on-screen overlay (news or interviews) 3. Last pass, go through the transcription and map voices+ids to actual names