TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Talk-Llama

474 pointsby plurbyover 1 year ago

20 comments

ggerganovover 1 year ago
Heh, funny to see this popup here :)<p>The performance on Apple Silicon should be much better today compared to what is shown in the video as whisper.cpp now runs fully on the GPU and there have been significant improvements in llama.cpp generation speed over the last few months.
评论 #38113936 未加载
评论 #38113405 未加载
评论 #38115967 未加载
评论 #38117816 未加载
评论 #38114039 未加载
评论 #38125049 未加载
评论 #38116315 未加载
wxnxover 1 year ago
This is cool. I hooked up Llama to an open-source TTS model for a recent project and there was lots of fun engineering that went into it.<p>On a different note:<p>I think the most useful coding copilot tools for me reduce &quot;manual overhead&quot; without attempting to do any hard thinking&#x2F;problem solving for me (such as generating arguments and types from docstrings or vice-versa, etc.). For more complicated tasks you really have to give copilot a pretty good &quot;starting point&quot;.<p>I often talk to myself while coding. It would be extremely, extremely futuristic (and potentially useful) if a tool like this embedded my speech into a context vector and used it to as an additional copilot input so the model has a better &quot;starting point&quot;.<p>I&#x27;m a late adopter of copilot and don&#x27;t use it all the time but if anyone is aware of anything like this I&#x27;d be curious to hear about it.
nwoliover 1 year ago
A few months until this is effectively outlawed if the open weights proposal in 270 days comes into existence
评论 #38113985 未加载
评论 #38113896 未加载
评论 #38117142 未加载
评论 #38114595 未加载
评论 #38112245 未加载
lbltavaresover 1 year ago
I&#x27;m getting a &quot;floating point exception&quot; when running .&#x2F;talk-llama on arch and debian. Already checked sdl2lib and ffmpeg (because of this issue: <a href="https:&#x2F;&#x2F;github.com&#x2F;ggerganov&#x2F;whisper.cpp&#x2F;issues&#x2F;1325">https:&#x2F;&#x2F;github.com&#x2F;ggerganov&#x2F;whisper.cpp&#x2F;issues&#x2F;1325</a>) but nothing seems to fix it. Anyone else?
评论 #38139300 未加载
avereveardover 1 year ago
Aren&#x27;t there text to talk solution that can receive a stream of text so one doesn&#x27;t have to wait for llama to finish production before getting the answer talked out?<p>I guess it&#x27;d only work if the model can keep the buffer filled fast enough so the tts engine doesn&#x27;t stall.
评论 #38111497 未加载
评论 #38111691 未加载
评论 #38113299 未加载
cjbprimeover 1 year ago
Would it be possible to reduce lag by streaming groups of ~6 tokens at a time to the TTS as they&#x27;re generated, instead of waiting for the full LLM response before beginning to speak it?
评论 #38117233 未加载
boiler_up800over 1 year ago
What’s the best chat interface for llama? I have a 3090 and would love to get one of the models running in my terminal for quick coding tasks.
评论 #38113458 未加载
评论 #38113116 未加载
评论 #38113650 未加载
评论 #38112266 未加载
评论 #38115965 未加载
d3nj4lover 1 year ago
This makes me wonder, what&#x27;s the equivalent to ollama for whisper&#x2F;SOTA OS tts models? I&#x27;m really happy with ollama for locally running OS LLMs, but I don&#x27;t know of any project that makes it <i>that</i> simple to set up whisper locally.
评论 #38113252 未加载
评论 #38114126 未加载
评论 #38113257 未加载
onemoresoopover 1 year ago
Could anyone explain the capability of this in plain English? Can this learn and retain context of a chat and build on some kind of long term memory? Thanks
评论 #38114048 未加载
kristopolousover 1 year ago
this has really strong eliza vibes.
SillyUsernameover 1 year ago
Does anybody have a quick start for building it all in Windows for this? I could probably check it out as a VS project and build but I&#x27;m going to bet since it&#x27;s not documented it&#x27;s going to have issues specifically because the Linux build instructions are the only ones that are a first class citizen...
human183729153over 1 year ago
What are currently the best&#x2F;go-to approaches to detect the end of an utterance? This can be tricky even in conversations between humans, requiring semantic information about what the other person is saying. I wonder if there’s any automated strategy that works well enough.
评论 #38144519 未加载
wahnfriedenover 1 year ago
How does this choose when to speak back? (Like is it after a pause, or other heuristics.) I tried looking through the source to find this logic.
评论 #38116544 未加载
rgbrgbover 1 year ago
very sick demo! if anyone wants to work on packaging this up for broader (swiftUI&#x2F;macos) consumption, I just added an issue <a href="https:&#x2F;&#x2F;github.com&#x2F;psugihara&#x2F;FreeChat&#x2F;issues&#x2F;30">https:&#x2F;&#x2F;github.com&#x2F;psugihara&#x2F;FreeChat&#x2F;issues&#x2F;30</a>
bribriover 1 year ago
Elevenlabs voice is amazing but it&#x27;s so expensive. You can easily spend $20 on a single conversation.
评论 #38122576 未加载
danielEMover 1 year ago
All I need is ... Vulkan support :) pls, pls, pls :)
columnover 1 year ago
why use this instead of &quot;memgpt run&quot; ?
评论 #38111773 未加载
评论 #38113682 未加载
评论 #38111480 未加载
horsellamaover 1 year ago
can this be used to talk with local documents?<p>say I have a research paper in pdf, can I ask llama questions about it?
评论 #38114114 未加载
BigRedDog1669over 1 year ago
Tacoma
birdyroosterover 1 year ago
I don’t want to talk to anything in my terminal. It’s a shitty interface for that.
评论 #38114666 未加载
评论 #38120195 未加载