{"slug": "pipecat", "name": "Pipecat", "category": "ai", "category_name": "AI models and assistants", "description": "Open-source Python framework for building real-time voice and multimodal conversational AI agents.", "recommenders": 6, "commenters": 11, "warners": 0, "score": 6.58, "rising": -0.64, "first": "2025-12-08", "latest": "2026-09-09", "why": "Liked as an open-source framework for building voice agents, where you can plug in almost any STT, LLM or TTS provider, including local models via Ollama or Nvidia models on a single GPU.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 6, "neutral": 5, "compare": 0, "warn": 0}, "spark": [0, 1, 2, 0, 0, 1, 0, 0, 1, 0, 1, 0], "listed": true, "loved": ["Pluggable providers", "Local models"], "caveat": "Needs tinkering", "short_quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "icon": true, "domain": "pipecat.ai", "github_repo": null, "official_url": "https://www.pipecat.ai", "pricing": "open source", "aliases": ["Pipecat"], "verdict": [{"text": "Liked as an open-source framework for building voice agents", "cites": [{"id": 785706449, "item_id": 49106653, "url": "https://news.ycombinator.com/item?id=49106653", "by": "teitoklien", "date": "2026-07-29", "stance": "recommend", "quote": "I literally have voice agents via claude agent sdk + Pipecat", "excerpt": "why ? I literally have voice agents via claude agent sdk + Pipecat , who i talk to the entire day whenever it suddenly handwaves at me with an indicator that it wants to talk to me, and it orchestrates and manages typically 6-7 git worktrees in parallel of my monorepo + coding and other non-coding work. Every. Single. Day. and reliably and robustly, to get cool stuff done. Why does it sound satire to you ? I was neve", "context": "", "use_case": "voice agent framework", "switched_from": "", "story": "Claude Is Down"}, {"id": 766666002, "item_id": 47916625, "url": "https://news.ycombinator.com/item?id=47916625", "by": "davemo", "date": "2026-04-26", "stance": "recommend", "quote": "since publishing we have swapped to Pipecat", "excerpt": "I love XState. It has solved so many problems in quite a few codebases for me, most recently it has become the backbone of a voice app I'm building for the banking sector. Thank you for all the time and effort you put into making it great. I blogged[0] a bit about my experience with Finite State Machines, and about the architecture we landed on with XState and Mastra (although since publishing we have swapped to Pipe", "context": "building a voice app for banking", "use_case": "voice agent framework", "switched_from": "Mastra", "story": "Statecharts: hierarchical state machines"}, {"id": 739038402, "item_id": 46189900, "url": "https://news.ycombinator.com/item?id=46189900", "by": "a6kme", "date": "2025-12-08", "stance": "neutral", "quote": "when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful", "excerpt": "Earlier I was using other platforms for production voice agents. One thing that became obvious was the cost: 60\u201370% of our total spend was the Vapi platform fee, and only 30-40% was actual LLM/STT/TTS usage. Platform cost dominated everything. That alone pushed us toward something self-hosted. But when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful an", "context": "", "use_case": "voice agents", "switched_from": "Vapi", "story": "Show HN: Dograh \u2013 an OSS Vapi alternative to quickly build and test voice agents"}]}, {"text": ", where you can plug in almost any STT, LLM or TTS provider, including local models via Ollama or Nvidia models on a single GPU.", "cites": [{"id": 747839504, "item_id": 46739969, "url": "https://news.ycombinator.com/item?id=46739969", "by": "dfajgljsldkjag", "date": "2026-01-23", "stance": "recommend", "quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "excerpt": "It requires a bit of tinkering, but I think pipecat is the way to go. You can plug in pretty much any STT/LLM/TTS you want and go. It definitely supports local models but its up to you to get your hands on those models. Not sure if there's any turnkey setups that are preconfigured for local install where you can just press play and go though. Last I heard E2E speech to speech models are still pretty weak. I've had pr", "context": "", "use_case": "local voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 740305296, "item_id": 46269081, "url": "https://news.ycombinator.com/item?id=46269081", "by": "kstonekuan", "date": "2025-12-14", "stance": "recommend", "quote": "Pipecat already supports that natively, so this can be done easily with ollama.", "excerpt": "Yes, Pipecat already supports that natively, so this can be done easily with ollama. I have also built that into the environment variables with `OLLAMA_BASE_URL`. About ollama in pipecat: https://docs.pipecat.ai/server/services/llm/ollama Also, check out any provider they support, and it can be easily onboarded in a few lines of code.", "context": "", "use_case": "voice dictation pipeline", "switched_from": "", "story": "Show HN: Open-source customizable AI voice dictation built on Pipecat"}, {"id": 747914016, "item_id": 46744626, "url": "https://news.ycombinator.com/item?id=46744626", "by": "supermatt", "date": "2026-01-24", "stance": "recommend", "quote": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat", "excerpt": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat Discussion: https://news.ycombinator.com/item?id=46528045 Article: https://www.daily.co/blog/building-voice-agents-with-nvidia-open-models/", "context": "", "use_case": "low-latency voice agents", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"text": " The most common caveat is that it takes tinkering to set up", "cites": [{"id": 747839504, "item_id": 46739969, "url": "https://news.ycombinator.com/item?id=46739969", "by": "dfajgljsldkjag", "date": "2026-01-23", "stance": "recommend", "quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "excerpt": "It requires a bit of tinkering, but I think pipecat is the way to go. You can plug in pretty much any STT/LLM/TTS you want and go. It definitely supports local models but its up to you to get your hands on those models. Not sure if there's any turnkey setups that are preconfigured for local install where you can just press play and go though. Last I heard E2E speech to speech models are still pretty weak. I've had pr", "context": "", "use_case": "local voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"text": ", and production plumbing like tracing and AI-to-AI test loops is still left to you.", "cites": [{"id": 739038402, "item_id": 46189900, "url": "https://news.ycombinator.com/item?id=46189900", "by": "a6kme", "date": "2025-12-08", "stance": "neutral", "quote": "when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful", "excerpt": "Earlier I was using other platforms for production voice agents. One thing that became obvious was the cost: 60\u201370% of our total spend was the Vapi platform fee, and only 30-40% was actual LLM/STT/TTS usage. Platform cost dominated everything. That alone pushed us toward something self-hosted. But when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful an", "context": "", "use_case": "voice agents", "switched_from": "Vapi", "story": "Show HN: Dograh \u2013 an OSS Vapi alternative to quickly build and test voice agents"}]}], "praise": [{"label": "Pluggable providers", "summary": "Commenters say you can swap in nearly any STT, LLM or TTS service with little code.", "quotes": [{"id": 747839504, "item_id": 46739969, "url": "https://news.ycombinator.com/item?id=46739969", "by": "dfajgljsldkjag", "date": "2026-01-23", "stance": "recommend", "quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "excerpt": "It requires a bit of tinkering, but I think pipecat is the way to go. You can plug in pretty much any STT/LLM/TTS you want and go. It definitely supports local models but its up to you to get your hands on those models. Not sure if there's any turnkey setups that are preconfigured for local install where you can just press play and go though. Last I heard E2E speech to speech models are still pretty weak. I've had pr", "context": "", "use_case": "local voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 740305296, "item_id": 46269081, "url": "https://news.ycombinator.com/item?id=46269081", "by": "kstonekuan", "date": "2025-12-14", "stance": "recommend", "quote": "Pipecat already supports that natively, so this can be done easily with ollama.", "excerpt": "Yes, Pipecat already supports that natively, so this can be done easily with ollama. I have also built that into the environment variables with `OLLAMA_BASE_URL`. About ollama in pipecat: https://docs.pipecat.ai/server/services/llm/ollama Also, check out any provider they support, and it can be easily onboarded in a few lines of code.", "context": "", "use_case": "voice dictation pipeline", "switched_from": "", "story": "Show HN: Open-source customizable AI voice dictation built on Pipecat"}, {"id": 747882675, "item_id": 46742667, "url": "https://news.ycombinator.com/item?id=46742667", "by": "nsbk", "date": "2026-01-24", "stance": "neutral", "quote": "pipecat to glue everything together", "excerpt": "I'm putting together a streaming ASR + LLM + streaming TTS setup based on Nvidia speech models: nemotron ASR and magpie TTS, pipecat to glue everything together, plus an LLM of your choice. I added Spanish support using canary models, as magpie models are English-only and it still works really well. The work is based on a repo by pipecat that I forked and modified to be more comfortable to run (docker compose for the", "context": "", "use_case": "voice pipeline glue", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"label": "Local models", "summary": "It supports local models, including Ollama and Nvidia models running on a single consumer GPU.", "quotes": [{"id": 740305296, "item_id": 46269081, "url": "https://news.ycombinator.com/item?id=46269081", "by": "kstonekuan", "date": "2025-12-14", "stance": "recommend", "quote": "Pipecat already supports that natively, so this can be done easily with ollama.", "excerpt": "Yes, Pipecat already supports that natively, so this can be done easily with ollama. I have also built that into the environment variables with `OLLAMA_BASE_URL`. About ollama in pipecat: https://docs.pipecat.ai/server/services/llm/ollama Also, check out any provider they support, and it can be easily onboarded in a few lines of code.", "context": "", "use_case": "voice dictation pipeline", "switched_from": "", "story": "Show HN: Open-source customizable AI voice dictation built on Pipecat"}, {"id": 747914016, "item_id": 46744626, "url": "https://news.ycombinator.com/item?id=46744626", "by": "supermatt", "date": "2026-01-24", "stance": "recommend", "quote": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat", "excerpt": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat Discussion: https://news.ycombinator.com/item?id=46528045 Article: https://www.daily.co/blog/building-voice-agents-with-nvidia-open-models/", "context": "", "use_case": "low-latency voice agents", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 747882675, "item_id": 46742667, "url": "https://news.ycombinator.com/item?id=46742667", "by": "nsbk", "date": "2026-01-24", "stance": "neutral", "quote": "pipecat to glue everything together", "excerpt": "I'm putting together a streaming ASR + LLM + streaming TTS setup based on Nvidia speech models: nemotron ASR and magpie TTS, pipecat to glue everything together, plus an LLM of your choice. I added Spanish support using canary models, as magpie models are English-only and it still works really well. The work is based on a repo by pipecat that I forked and modified to be more comfortable to run (docker compose for the", "context": "", "use_case": "voice pipeline glue", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"label": "Real voice agents", "summary": "Commenters use it for working voice agents, from daily coding assistants to a banking voice app.", "quotes": [{"id": 785706449, "item_id": 49106653, "url": "https://news.ycombinator.com/item?id=49106653", "by": "teitoklien", "date": "2026-07-29", "stance": "recommend", "quote": "I literally have voice agents via claude agent sdk + Pipecat", "excerpt": "why ? I literally have voice agents via claude agent sdk + Pipecat , who i talk to the entire day whenever it suddenly handwaves at me with an indicator that it wants to talk to me, and it orchestrates and manages typically 6-7 git worktrees in parallel of my monorepo + coding and other non-coding work. Every. Single. Day. and reliably and robustly, to get cool stuff done. Why does it sound satire to you ? I was neve", "context": "", "use_case": "voice agent framework", "switched_from": "", "story": "Claude Is Down"}, {"id": 766666002, "item_id": 47916625, "url": "https://news.ycombinator.com/item?id=47916625", "by": "davemo", "date": "2026-04-26", "stance": "recommend", "quote": "since publishing we have swapped to Pipecat", "excerpt": "I love XState. It has solved so many problems in quite a few codebases for me, most recently it has become the backbone of a voice app I'm building for the banking sector. Thank you for all the time and effort you put into making it great. I blogged[0] a bit about my experience with Finite State Machines, and about the architecture we landed on with XState and Mastra (although since publishing we have swapped to Pipe", "context": "building a voice app for banking", "use_case": "voice agent framework", "switched_from": "Mastra", "story": "Statecharts: hierarchical state machines"}, {"id": 753028371, "item_id": 47064273, "url": "https://news.ycombinator.com/item?id=47064273", "by": "Blair_Erickson", "date": "2026-02-18", "stance": "neutral", "quote": "Pipecat voice pipeline", "excerpt": "I'm Blair, one of the creators of PearlOS. Here's what we've built. PearlOS is a self-evolving operating system. You talk to her, she builds, creates, informs, learns, evolves. This is an open source browser-based, voice and touch desktop environment powered by an intelligent learning companion named Pearl. Pearl has full awareness and control of your visual and audio experience inside PearlOS in real time, so you ca", "context": "", "use_case": "voice pipeline", "switched_from": "", "story": "Show HN: PearlOS \u2013An open source OS companion that learns and evolves around you"}]}, {"label": "Low latency", "summary": "One commenter points to a demo of low-latency end-to-end voice using Nvidia models on one GPU.", "quotes": [{"id": 747914016, "item_id": 46744626, "url": "https://news.ycombinator.com/item?id=46744626", "by": "supermatt", "date": "2026-01-24", "stance": "recommend", "quote": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat", "excerpt": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat Discussion: https://news.ycombinator.com/item?id=46528045 Article: https://www.daily.co/blog/building-voice-agents-with-nvidia-open-models/", "context": "", "use_case": "low-latency voice agents", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"label": "Active demos", "summary": "The team is noted for frequently putting out interesting demos.", "quotes": [{"id": 794066832, "item_id": 49629177, "url": "https://news.ycombinator.com/item?id=49629177", "by": "arach", "date": "2026-09-09", "stance": "recommend", "quote": "Are you familiar with Pipecat https://github.com/pipecat-ai/pipecat - they put out interesting demos frequently", "excerpt": "I think we're close to the point where local models can make this economically viable as an alternative to hands off XAI/OAI/ANT You need rapid good enough rapid turn response + rapid good enough TTS in order to make it economically for someone who's not blending it into a more compute heavy pricing model Have you looked into it? Are you familiar with Pipecat https://github.com/pipecat-ai/pipecat - they put out inter", "context": "", "use_case": "voice agents", "switched_from": "", "story": "Ask HN: Anyone using dictation with coding agents?"}]}], "caveats": [{"label": "Needs tinkering", "summary": "Setup takes some tinkering, and no turnkey preconfigured local install is known to the commenter.", "quotes": [{"id": 747839504, "item_id": 46739969, "url": "https://news.ycombinator.com/item?id=46739969", "by": "dfajgljsldkjag", "date": "2026-01-23", "stance": "recommend", "quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "excerpt": "It requires a bit of tinkering, but I think pipecat is the way to go. You can plug in pretty much any STT/LLM/TTS you want and go. It definitely supports local models but its up to you to get your hands on those models. Not sure if there's any turnkey setups that are preconfigured for local install where you can just press play and go though. Last I heard E2E speech to speech models are still pretty weak. I've had pr", "context": "", "use_case": "local voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}, {"label": "Missing plumbing", "summary": "One team found variable extraction, LLM call tracing and AI-to-AI testing still painful to build.", "quotes": [{"id": 739038402, "item_id": 46189900, "url": "https://news.ycombinator.com/item?id=46189900", "by": "a6kme", "date": "2025-12-08", "stance": "neutral", "quote": "when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful", "excerpt": "Earlier I was using other platforms for production voice agents. One thing that became obvious was the cost: 60\u201370% of our total spend was the Vapi platform fee, and only 30-40% was actual LLM/STT/TTS usage. Platform cost dominated everything. That alone pushed us toward something self-hosted. But when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful an", "context": "", "use_case": "voice agents", "switched_from": "Vapi", "story": "Show HN: Dograh \u2013 an OSS Vapi alternative to quickly build and test voice agents"}]}, {"label": "Speech-to-speech", "summary": "One commenter has not yet gotten local speech-to-speech setups working with it.", "quotes": [{"id": 747868976, "item_id": 46741811, "url": "https://news.ycombinator.com/item?id=46741811", "by": "dsrtslnd23", "date": "2026-01-23", "stance": "neutral", "quote": "I am currently playing with pipecat", "excerpt": "yes, I am currently playing with pipecat - both with ASR + LLM + TTS pipeline and also speech to text (ultravox) + TTS but haven't been successful with local speech to speech setups yet.", "context": "", "use_case": "voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}]}], "who": [{"by": "davemo", "context": "building a voice app for banking", "url": "https://news.ycombinator.com/item?id=47916625", "date": "2026-04-26"}], "tags": ["for voice agents", "for local models", "self-hostable"], "use_cases": ["voice agents", "voice agent framework", "customer-facing voice agents", "voice pipeline", "low-latency voice agents", "voice pipeline glue"], "compared": [{"slug": "mastra", "name": "Mastra", "co": 1, "switched_from": 1, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=46731068", "title": "Ask HN: What's the current best local/open speech-to-speech setup?", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=49610888", "title": "Ask HN: Anyone using dictation with coding agents?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49548600", "title": "Ask HN: Who is using MCP in production?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49102150", "title": "Claude Is Down", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47908833", "title": "Statecharts: hierarchical state machines", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47064244", "title": "Show HN: PearlOS \u2013An open source OS companion that learns and evolves around you", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {"neutral": 1, "recommend": 1}, "2026-01": {"recommend": 2, "neutral": 2}, "2026-02": {"neutral": 1}, "2026-03": {}, "2026-04": {"recommend": 1}, "2026-05": {}, "2026-06": {}, "2026-07": {"recommend": 1}, "2026-08": {}, "2026-09": {"neutral": 1, "recommend": 1}, "2026-10": {}}, "quotes": [{"id": 794066832, "item_id": 49629177, "url": "https://news.ycombinator.com/item?id=49629177", "by": "arach", "date": "2026-09-09", "stance": "recommend", "quote": "Are you familiar with Pipecat https://github.com/pipecat-ai/pipecat - they put out interesting demos frequently", "excerpt": "I think we're close to the point where local models can make this economically viable as an alternative to hands off XAI/OAI/ANT You need rapid good enough rapid turn response + rapid good enough TTS in order to make it economically for someone who's not blending it into a more compute heavy pricing model Have you looked into it? Are you familiar with Pipecat https://github.com/pipecat-ai/pipecat - they put out inter", "context": "", "use_case": "voice agents", "switched_from": "", "story": "Ask HN: Anyone using dictation with coding agents?"}, {"id": 792961906, "item_id": 49560119, "url": "https://news.ycombinator.com/item?id=49560119", "by": "MitziMoto", "date": "2026-09-04", "stance": "neutral", "quote": "point any voice agent platform we choose -- eleven labs, vapi, pipecat, whatever", "excerpt": "We use MCP in production for our customer facing voice agents. Our custom MCP server defines tools and resources that the voice agents need access to to interact with our customers. (E.g. scheduling appointments, checking order status, etc). Now we can point any voice agent platform we choose -- eleven labs, vapi, pipecat, whatever -- at our custom MCP server and it instantly has an understanding of the tools availab", "context": "", "use_case": "customer-facing voice agents", "switched_from": "", "story": "Ask HN: Who is using MCP in production?"}, {"id": 785706449, "item_id": 49106653, "url": "https://news.ycombinator.com/item?id=49106653", "by": "teitoklien", "date": "2026-07-29", "stance": "recommend", "quote": "I literally have voice agents via claude agent sdk + Pipecat", "excerpt": "why ? I literally have voice agents via claude agent sdk + Pipecat , who i talk to the entire day whenever it suddenly handwaves at me with an indicator that it wants to talk to me, and it orchestrates and manages typically 6-7 git worktrees in parallel of my monorepo + coding and other non-coding work. Every. Single. Day. and reliably and robustly, to get cool stuff done. Why does it sound satire to you ? I was neve", "context": "", "use_case": "voice agent framework", "switched_from": "", "story": "Claude Is Down"}, {"id": 766666002, "item_id": 47916625, "url": "https://news.ycombinator.com/item?id=47916625", "by": "davemo", "date": "2026-04-26", "stance": "recommend", "quote": "since publishing we have swapped to Pipecat", "excerpt": "I love XState. It has solved so many problems in quite a few codebases for me, most recently it has become the backbone of a voice app I'm building for the banking sector. Thank you for all the time and effort you put into making it great. I blogged[0] a bit about my experience with Finite State Machines, and about the architecture we landed on with XState and Mastra (although since publishing we have swapped to Pipe", "context": "building a voice app for banking", "use_case": "voice agent framework", "switched_from": "Mastra", "story": "Statecharts: hierarchical state machines"}, {"id": 753028371, "item_id": 47064273, "url": "https://news.ycombinator.com/item?id=47064273", "by": "Blair_Erickson", "date": "2026-02-18", "stance": "neutral", "quote": "Pipecat voice pipeline", "excerpt": "I'm Blair, one of the creators of PearlOS. Here's what we've built. PearlOS is a self-evolving operating system. You talk to her, she builds, creates, informs, learns, evolves. This is an open source browser-based, voice and touch desktop environment powered by an intelligent learning companion named Pearl. Pearl has full awareness and control of your visual and audio experience inside PearlOS in real time, so you ca", "context": "", "use_case": "voice pipeline", "switched_from": "", "story": "Show HN: PearlOS \u2013An open source OS companion that learns and evolves around you"}, {"id": 747914016, "item_id": 46744626, "url": "https://news.ycombinator.com/item?id=46744626", "by": "supermatt", "date": "2026-01-24", "stance": "recommend", "quote": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat", "excerpt": "There was a great post the other day showing low latency end to end using Nvidia models on a single GPU with pipecat Discussion: https://news.ycombinator.com/item?id=46528045 Article: https://www.daily.co/blog/building-voice-agents-with-nvidia-open-models/", "context": "", "use_case": "low-latency voice agents", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 747882675, "item_id": 46742667, "url": "https://news.ycombinator.com/item?id=46742667", "by": "nsbk", "date": "2026-01-24", "stance": "neutral", "quote": "pipecat to glue everything together", "excerpt": "I'm putting together a streaming ASR + LLM + streaming TTS setup based on Nvidia speech models: nemotron ASR and magpie TTS, pipecat to glue everything together, plus an LLM of your choice. I added Spanish support using canary models, as magpie models are English-only and it still works really well. The work is based on a repo by pipecat that I forked and modified to be more comfortable to run (docker compose for the", "context": "", "use_case": "voice pipeline glue", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 747868976, "item_id": 46741811, "url": "https://news.ycombinator.com/item?id=46741811", "by": "dsrtslnd23", "date": "2026-01-23", "stance": "neutral", "quote": "I am currently playing with pipecat", "excerpt": "yes, I am currently playing with pipecat - both with ASR + LLM + TTS pipeline and also speech to text (ultravox) + TTS but haven't been successful with local speech to speech setups yet.", "context": "", "use_case": "voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 747839504, "item_id": 46739969, "url": "https://news.ycombinator.com/item?id=46739969", "by": "dfajgljsldkjag", "date": "2026-01-23", "stance": "recommend", "quote": "I think pipecat is the way to go.  You can plug in pretty much any STT/LLM/TTS you want and go.", "excerpt": "It requires a bit of tinkering, but I think pipecat is the way to go. You can plug in pretty much any STT/LLM/TTS you want and go. It definitely supports local models but its up to you to get your hands on those models. Not sure if there's any turnkey setups that are preconfigured for local install where you can just press play and go though. Last I heard E2E speech to speech models are still pretty weak. I've had pr", "context": "", "use_case": "local voice agent pipeline", "switched_from": "", "story": "Ask HN: What's the current best local/open speech-to-speech setup?"}, {"id": 740305296, "item_id": 46269081, "url": "https://news.ycombinator.com/item?id=46269081", "by": "kstonekuan", "date": "2025-12-14", "stance": "recommend", "quote": "Pipecat already supports that natively, so this can be done easily with ollama.", "excerpt": "Yes, Pipecat already supports that natively, so this can be done easily with ollama. I have also built that into the environment variables with `OLLAMA_BASE_URL`. About ollama in pipecat: https://docs.pipecat.ai/server/services/llm/ollama Also, check out any provider they support, and it can be easily onboarded in a few lines of code.", "context": "", "use_case": "voice dictation pipeline", "switched_from": "", "story": "Show HN: Open-source customizable AI voice dictation built on Pipecat"}, {"id": 739038402, "item_id": 46189900, "url": "https://news.ycombinator.com/item?id=46189900", "by": "a6kme", "date": "2025-12-08", "stance": "neutral", "quote": "when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful", "excerpt": "Earlier I was using other platforms for production voice agents. One thing that became obvious was the cost: 60\u201370% of our total spend was the Vapi platform fee, and only 30-40% was actual LLM/STT/TTS usage. Platform cost dominated everything. That alone pushed us toward something self-hosted. But when we switched to OSS stacks (Pipecat, LiveKit), we realise that even with great OSS, the plumbing was still painful an", "context": "", "use_case": "voice agents", "switched_from": "Vapi", "story": "Show HN: Dograh \u2013 an OSS Vapi alternative to quickly build and test voice agents"}]}