for reasoning it less intelligent model we need fallbacks to more intelligent models with good routing"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Jev introduces a new shape of LLM"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://simonwillison.net/2026/Sep/21/jev/"}},"_tags":["comment","author_iharnoor","story_49796843"],"author":"iharnoor","comment_text":"I am hyped to use Jev with AssemblyAI for voice agents and TTS but you can't rely on Jev for high reasoning tasks at all.
for reasoning it less intelligent model we need fallbacks to more intelligent models with good routing","created_at":"2026-09-24T18:40:59Z","created_at_i":1790275259,"objectID":"49835059","parent_id":49797803,"story_id":49796843,"story_title":"Jev introduces a new shape of LLM","story_url":"https://simonwillison.net/2026/Sep/21/jev/","updated_at":"2026-09-24T18:41:53Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Today, Universal-3.5 Pro is available on OpenRouter, served through our Sync Speech-to-Text API.
If you're already routing your LLM calls through OpenRouter, the speech-to-text call at the front of your pipeline now runs on the same key.
50% off list through OpenRouter for the first week: https://www.assemblyai.com/blog/aai-on-openrouter"},"story_title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"OpenRouter x AssemblyAI: Universal\u20113.5 Pro is now available on OpenRouter"},"story_url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"https://www.assemblyai.com/blog/aai-on-openrouter"}},"_tags":["comment","author_iharnoor","story_49822936"],"author":"iharnoor","comment_text":"Today, Universal-3.5 Pro is available on OpenRouter, served through our Sync Speech-to-Text API.
If you're already routing your LLM calls through OpenRouter, the speech-to-text call at the front of your pipeline now runs on the same key.
50% off list through OpenRouter for the first week: https://www.assemblyai.com/blog/aai-on-openrouter","created_at":"2026-09-23T21:39:08Z","created_at_i":1790199548,"objectID":"49822943","parent_id":49822936,"story_id":49822936,"story_title":"OpenRouter x AssemblyAI: Universal\u20113.5 Pro is now available on OpenRouter","story_url":"https://www.assemblyai.com/blog/aai-on-openrouter","updated_at":"2026-09-23T21:42:36Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Same here, I work for AssemblyAI and still encounter the same problem. When I call a dentist, they tell me to check with my health insurance and ask me to add us to the portal. Then I have to call the dentist again to make an appointment.
This entire cycle should be automated soon."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"There's a 100% Chance AI Agents Are Ruining the Internet"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://www.404media.co/theres-a-100-chance-ai-agents-are-already-ruining-the-internet/"}},"_tags":["comment","author_iharnoor","story_49715113"],"author":"iharnoor","comment_text":"Same here, I work for AssemblyAI and still encounter the same problem. When I call a dentist, they tell me to check with my health insurance and ask me to add us to the portal. Then I have to call the dentist again to make an appointment.
This entire cycle should be automated soon.","created_at":"2026-09-18T22:16:26Z","created_at_i":1789769786,"objectID":"49760976","parent_id":49715637,"story_id":49715113,"story_title":"There's a 100% Chance AI Agents Are Ruining the Internet","story_url":"https://www.404media.co/theres-a-100-chance-ai-agents-are-already-ruining-the-internet/","updated_at":"2026-09-18T22:18:36Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"lol same for me, I work for AssemblyAI
and still see when I call Dentist, they tell me check with health insurance and ask them add us to portal then I call Dentist again and then make an appointment.
This whole cycle needs to be automated and will be soon!!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"There's a 100% Chance AI Agents Are Ruining the Internet"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://www.404media.co/theres-a-100-chance-ai-agents-are-already-ruining-the-internet/"}},"_tags":["comment","author_iharnoor","story_49715113"],"author":"iharnoor","comment_text":"lol same for me, I work for AssemblyAI
and still see when I call Dentist, they tell me check with health insurance and ask them add us to portal then I call Dentist again and then make an appointment.
This whole cycle needs to be automated and will be soon!!","created_at":"2026-09-18T22:04:11Z","created_at_i":1789769051,"objectID":"49760862","parent_id":49715971,"story_id":49715113,"story_title":"There's a 100% Chance AI Agents Are Ruining the Internet","story_url":"https://www.404media.co/theres-a-100-chance-ai-agents-are-already-ruining-the-internet/","updated_at":"2026-09-18T22:08:05Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Today we're launching the Dictation API from AssemblyAI, the first ever API built for dictation: your users speak, and it returns a response tailored to your application without any user editing required.
Here's a short clip through our Transcription API versus through the new Dictation API:
- "um so can we uh move the the meeting to thursday i think friday works better actually"\n- "Can we move the meeting to Friday? That works better."
Traditional transcription models are trained to be verbatim and capture everything you say.
Dictation and Voice Input models need to be aligned for a different task: a cleaned up and properly formatted input to your system or application.
- Voice typing emails and slack messages.\n- Doctors dictating chart notes.\n- Voice input to robots and hardware devices.\n- Coding with your voice.
These are tasks our Dictation API is built for.
We've seen a surge in demand from developers building these types of applications, and the Dictation API now makes it much easier.
It's a simple, low-cost API any developer can use to ship dictation or voice input features and apps within minutes.
- Responses in <200ms\n- Built on our Universal-3.5 Pro transcription model, ranked #1 for accuracy on independent benchmarks.\n- Transcribes 30+ languages out of the box and supports language switching.\n- Include custom formatting instructions on the fly\n- $0.62 per hour of audio, all in - no token math.
We've also open\u2011sourced a free Mac dictation app built on the new Dictation API, so you can experience its speed and accuracy firsthand. It's called Blurt."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"The first ever API built for dictation;Today we're launching the Dictation API"},"story_url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"https://www.assemblyai.com/dashboard/login"}},"_tags":["comment","author_iharnoor","story_49722557"],"author":"iharnoor","comment_text":"Today we're launching the Dictation API from AssemblyAI, the first ever API built for dictation: your users speak, and it returns a response tailored to your application without any user editing required.
Here's a short clip through our Transcription API versus through the new Dictation API:
- "um so can we uh move the the meeting to thursday i think friday works better actually"\n- "Can we move the meeting to Friday? That works better."
Traditional transcription models are trained to be verbatim and capture everything you say.
Dictation and Voice Input models need to be aligned for a different task: a cleaned up and properly formatted input to your system or application.
- Voice typing emails and slack messages.\n- Doctors dictating chart notes.\n- Voice input to robots and hardware devices.\n- Coding with your voice.
These are tasks our Dictation API is built for.
We've seen a surge in demand from developers building these types of applications, and the Dictation API now makes it much easier.
It's a simple, low-cost API any developer can use to ship dictation or voice input features and apps within minutes.
- Responses in <200ms\n- Built on our Universal-3.5 Pro transcription model, ranked #1 for accuracy on independent benchmarks.\n- Transcribes 30+ languages out of the box and supports language switching.\n- Include custom formatting instructions on the fly\n- $0.62 per hour of audio, all in - no token math.
We've also open\u2011sourced a free Mac dictation app built on the new Dictation API, so you can experience its speed and accuracy firsthand. It's called Blurt.","created_at":"2026-09-16T05:50:05Z","created_at_i":1789537805,"objectID":"49722559","parent_id":49722557,"story_id":49722557,"story_title":"The first ever API built for dictation;Today we're launching the Dictation API","story_url":"https://www.assemblyai.com/dashboard/login","updated_at":"2026-09-16T05:52:56Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"mikodin"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Greetings friends! I have been joyfully working on a native iOS app called Sanctum. I really built this for myself and use it multiple times a day (https://apps.apple.com/us/app/sanctum-the-voice-journal/id67...)
Sanctum is a voice journal, think Obsidian but for voice notes. I am really honing in on data sovereignty, privacy, 1 click to record and self organization of notes / ideas.
While this isn't anything revolutionary, or technically difficult, I do think I've landed on a form factor and stack that is in my opinion not served by current products. So much so that I just simply had to build it myself. Sanctum was born out of an Obsidian plugin that I built and have been maintaining for about 2 years called Scribe and it's been a wonderful way to really hone in on a more full fledged agentic development process for a full stack experience.
The stack is simple
- For transcription I am using AssemblyAI - they have solid terms, 24hr retention and no training\n\n - LLM inference is running inside a TEE using redpill.ai, it's slow af but as private as I could find.\n\n - For cloud backup I am using full E2E encryption with Backblaze for storage\n\n - Endpoints and the like are AWS lambdas, DynamoDB for billing storage, Supabase for auth\nAnyway, something I am most definitely curious on is if people are recording their thoughts, what they are using (either a cobbled together solution or a product) and what value they feel they've been getting from it!"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: What are you working on? (September 2026)"}},"_tags":["comment","author_mikodin","story_49686380"],"author":"mikodin","comment_text":"Greetings friends! I have been joyfully working on a native iOS app called Sanctum. I really built this for myself and use it multiple times a day (https://apps.apple.com/us/app/sanctum-the-voice-journal/id67...)
Sanctum is a voice journal, think Obsidian but for voice notes. I am really honing in on data sovereignty, privacy, 1 click to record and self organization of notes / ideas.
While this isn't anything revolutionary, or technically difficult, I do think I've landed on a form factor and stack that is in my opinion not served by current products. So much so that I just simply had to build it myself. Sanctum was born out of an Obsidian plugin that I built and have been maintaining for about 2 years called Scribe and it's been a wonderful way to really hone in on a more full fledged agentic development process for a full stack experience.
The stack is simple
- For transcription I am using AssemblyAI - they have solid terms, 24hr retention and no training\n\n - LLM inference is running inside a TEE using redpill.ai, it's slow af but as private as I could find.\n\n - For cloud backup I am using full E2E encryption with Backblaze for storage\n\n - Endpoints and the like are AWS lambdas, DynamoDB for billing storage, Supabase for auth\nAnyway, something I am most definitely curious on is if people are recording their thoughts, what they are using (either a cobbled together solution or a product) and what value they feel they've been getting from it!","created_at":"2026-09-14T23:00:57Z","created_at_i":1789426857,"objectID":"49705436","parent_id":49686380,"story_id":49686380,"story_title":"Ask HN: What are you working on? (September 2026)","updated_at":"2026-09-14T23:03:38Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"I still feel on device Dication is not accurate yet.
Coming here as Developer Relations at AssemblyAI, specially where entities need to be maintained such as names from different cultures.
Can we try comparison by switching API call to our Dictation API."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: SayItFlow \u2013 On-device Mac dictation that unloads models when idle"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/innovatorved/sayItflow"}},"_tags":["comment","author_iharnoor","story_49701374"],"author":"iharnoor","comment_text":"I still feel on device Dication is not accurate yet.
Coming here as Developer Relations at AssemblyAI, specially where entities need to be maintained such as names from different cultures.
Can we try comparison by switching API call to our Dictation API.","created_at":"2026-09-14T19:45:18Z","created_at_i":1789415118,"objectID":"49702814","parent_id":49701374,"story_id":49701374,"story_title":"Show HN: SayItFlow \u2013 On-device Mac dictation that unloads models when idle","story_url":"https://github.com/innovatorved/sayItflow","updated_at":"2026-09-14T19:46:06Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"By next month the competition for TTS will be even more!
Voice models are not winner take all market unlike LLM APIs
Coming here as Developer Relations at AssemblyAI"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Nari Qwen3-TTS and Qwen3-ASR \u2013 High accuracy, low latency and cost"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://narilabs.com/blog/nari-labs-leads-coval-voice-ai-benchmarks/"}},"_tags":["comment","author_iharnoor","story_49699267"],"author":"iharnoor","comment_text":"By next month the competition for TTS will be even more!
Voice models are not winner take all market unlike LLM APIs
Coming here as Developer Relations at AssemblyAI","created_at":"2026-09-14T19:41:15Z","created_at_i":1789414875,"objectID":"49702762","parent_id":49699267,"story_id":49699267,"story_title":"Show HN: Nari Qwen3-TTS and Qwen3-ASR \u2013 High accuracy, low latency and cost","story_url":"https://narilabs.com/blog/nari-labs-leads-coval-voice-ai-benchmarks/","updated_at":"2026-09-15T02:40:23Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"iharnoor"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"I was surprised Apple talked about note taking like Granola for Apple watch but didn't integrate that or talk about for iPhone Duo.
btw. we power Granola, I work at AssemblyAI"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"iPhone Duo"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://www.apple.com/iphone-duo/"}},"_tags":["comment","author_iharnoor","story_49630931"],"author":"iharnoor","comment_text":"I was surprised Apple talked about note taking like Granola for Apple watch but didn't integrate that or talk about for iPhone Duo.
btw. we power Granola, I work at AssemblyAI","created_at":"2026-09-12T01:31:13Z","created_at_i":1789176673,"objectID":"49667721","parent_id":49630931,"story_id":49630931,"story_title":"iPhone Duo","story_url":"https://www.apple.com/iphone-duo/","updated_at":"2026-09-12T01:34:44Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"473999"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Obi9 | Full-Stack Engineer | REMOTE (US time zones) | Contract or Full-time | https://obi9.ai/jobs/full-stack-engineer
A hedge fund analyst covering therapeutics has to watch hundreds of companies at once: press releases, SEC filings, management calls, updates on government sites, clinical data publications. Obi9 ingests all of it and surface what matters. Hedge funds pay for it today.
We are early and the team is small, so you would own entire surfaces of the product rather than pick up tickets:
- Ship full-stack features end to end, from the data model to the interface.
- Build the ingestion and enrichment pipelines that turn raw data into rich insights and alerts.
- Work on AI-backed features, from ideation to serving them in the webapp and over MCP.
- Make heavy data legible: dense tables, financial charts, document viewers, a live alert stream.
Stack: React + TypeScript, Node, Elixir, Python, distributed AI pipelines.
Beyond full-stack ability we want someone who already understands one of these worlds. One is enough, we are not asking for all four:
- Financial information systems. Bloomberg, FactSet, AlphaSense, Capital IQ, PitchBook. You know what entity resolution, identifier mapping and point-in-time data cost to get right.
- Financial domain knowledge in healthcare. Equity research, valuation, or time spent living inside filings and models.
- Audio and transcription. OtterAI Deepgram, AssemblyAI, Whisper. Diarization, speaker attribution, turning messy call audio into something quotable.
- AI-intensive products where the pipeline IS the product. Legal AI, document intelligence, medical coding. You know what evals, grounding and extraction schemas that hold actually cost.
Minimum: 5+ years building production web applications, a technical degree, fluency with current AI tooling, and eight continuous hours inside the 8:30am to 5:00pm ET window.
Apply: https://obi9.ai/jobs/full-stack-engineer?tab=apply"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Who is hiring? (September 2026)"}},"_tags":["comment","author_473999","story_49522897"],"author":"473999","children":[49718752],"comment_text":"Obi9 | Full-Stack Engineer | REMOTE (US time zones) | Contract or Full-time | https://obi9.ai/jobs/full-stack-engineer
A hedge fund analyst covering therapeutics has to watch hundreds of companies at once: press releases, SEC filings, management calls, updates on government sites, clinical data publications. Obi9 ingests all of it and surface what matters. Hedge funds pay for it today.
We are early and the team is small, so you would own entire surfaces of the product rather than pick up tickets:
- Ship full-stack features end to end, from the data model to the interface.
- Build the ingestion and enrichment pipelines that turn raw data into rich insights and alerts.
- Work on AI-backed features, from ideation to serving them in the webapp and over MCP.
- Make heavy data legible: dense tables, financial charts, document viewers, a live alert stream.
Stack: React + TypeScript, Node, Elixir, Python, distributed AI pipelines.
Beyond full-stack ability we want someone who already understands one of these worlds. One is enough, we are not asking for all four:
- Financial information systems. Bloomberg, FactSet, AlphaSense, Capital IQ, PitchBook. You know what entity resolution, identifier mapping and point-in-time data cost to get right.
- Financial domain knowledge in healthcare. Equity research, valuation, or time spent living inside filings and models.
- Audio and transcription. OtterAI Deepgram, AssemblyAI, Whisper. Diarization, speaker attribution, turning messy call audio into something quotable.
- AI-intensive products where the pipeline IS the product. Legal AI, document intelligence, medical coding. You know what evals, grounding and extraction schemas that hold actually cost.
Minimum: 5+ years building production web applications, a technical degree, fluency with current AI tooling, and eight continuous hours inside the 8:30am to 5:00pm ET window.
Apply: https://obi9.ai/jobs/full-stack-engineer?tab=apply","created_at":"2026-09-11T16:57:55Z","created_at_i":1789145875,"objectID":"49661611","parent_id":49522897,"story_id":49522897,"story_title":"Ask HN: Who is hiring? (September 2026)","updated_at":"2026-09-15T20:52:41Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"andai"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"I get the auto-transcript with yt-dlp then ask a cheap LLM like DeepSeek to clean it up.
Though lately I've been uploading the audio to AssemblyAI, I somehow still haven't used up my credits after several years lol
At one point I built a system that would summarize the transcript and I'd be able to ask questions about it, but Gemini can do that natively now so I usually just use that."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Casey Muratori \u2013 The Root of the Root of All Evil \u2013 BSC 2026 [video]"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://www.youtube.com/watch?v=hpj6r6CjJf8"}},"_tags":["comment","author_andai","story_49463888"],"author":"andai","comment_text":"I get the auto-transcript with yt-dlp then ask a cheap LLM like DeepSeek to clean it up.
Though lately I've been uploading the audio to AssemblyAI, I somehow still haven't used up my credits after several years lol
At one point I built a system that would summarize the transcript and I'd be able to ask questions about it, but Gemini can do that natively now so I usually just use that.","created_at":"2026-08-30T14:38:23Z","created_at_i":1788100703,"objectID":"49499066","parent_id":49498939,"story_id":49463888,"story_title":"Casey Muratori \u2013 The Root of the Root of All Evil \u2013 BSC 2026 [video]","story_url":"https://www.youtube.com/watch?v=hpj6r6CjJf8","updated_at":"2026-08-30T22:17:06Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"andai"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"So I tried this with GLM 5.3 and it kinda worked. I used AssemblyAI cause I don't have the ability to run local models.
The result is pretty ugly and unintuitive, OP's design is much nicer. And there are desync issues, but it's not too bad for a one-shot. (Kinda crazy that we can even do that these days, and then complain about it instead of being amazed!)"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Vocal Slice \u2013 Cut audio by selecting text, fully on-device"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://vocalslice.com/"}},"_tags":["comment","author_andai","story_49245021"],"author":"andai","comment_text":"So I tried this with GLM 5.3 and it kinda worked. I used AssemblyAI cause I don't have the ability to run local models.
The result is pretty ugly and unintuitive, OP's design is much nicer. And there are desync issues, but it's not too bad for a one-shot. (Kinda crazy that we can even do that these days, and then complain about it instead of being amazed!)","created_at":"2026-08-18T18:50:19Z","created_at_i":1787079019,"objectID":"49350681","parent_id":49341703,"story_id":49245021,"story_title":"Show HN: Vocal Slice \u2013 Cut audio by selecting text, fully on-device","story_url":"https://vocalslice.com/","updated_at":"2026-08-18T18:55:38Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"robinyapockets"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Hey HN, I built Ellis, an AI notetaker for individuals, not orgs. And since my personal schedule includes meetings outside of Zoom, it was important that it works well for in-person meetings.
In fact to me, in-person conversations have always been a greater source of learning and creative ideas than virtual. On Zoom, the fixed screen and talking points tend to bring out a more transactional nature in me.
Yet building a notetaker for in-person isn't trivial. Online, each speaker has their own audio stream. In a room (or during a walk outside) everyone shares the same mic.
Models like AssemblyAI give you transcriptions with Speaker A and Speaker B, but you still have to mark yourself as the one who said X and not Y.
This is especially true if I want to get coached from my conversations. A useful use case for me as an introvert.
Here's how Ellis does this:\n1) You create a voice profile of yourself (Deepgram diarization + pyannote's wespeaker model)\n2) It splits the recording into speaker segments using AssemblyAI\n3) It surfaces the most distinctive line from each unnamed speaker \u2014 "Moana is the best movie ever!"\n4) You tap to confirm.
Remember to ask for consent! Ellis prompts you with ways to ask at the start of a recording.
Still not solved: two near-identical voices in the same room (e.g. me and my dad). Curious if anyone found a diarization approach that separates near-identical voices without per-speaker enrollment?
If you have an opinion about the tool, I'm all ears! You can try it for free."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Cleaner speaker identification for in-person meeting transcripts"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://apps.apple.com/us/app/ellis-in-person-ai-notes/id6753657763"}},"_tags":["comment","author_robinyapockets","story_49208829"],"author":"robinyapockets","comment_text":"Hey HN, I built Ellis, an AI notetaker for individuals, not orgs. And since my personal schedule includes meetings outside of Zoom, it was important that it works well for in-person meetings.
In fact to me, in-person conversations have always been a greater source of learning and creative ideas than virtual. On Zoom, the fixed screen and talking points tend to bring out a more transactional nature in me.
Yet building a notetaker for in-person isn't trivial. Online, each speaker has their own audio stream. In a room (or during a walk outside) everyone shares the same mic.
Models like AssemblyAI give you transcriptions with Speaker A and Speaker B, but you still have to mark yourself as the one who said X and not Y.
This is especially true if I want to get coached from my conversations. A useful use case for me as an introvert.
Here's how Ellis does this:\n1) You create a voice profile of yourself (Deepgram diarization + pyannote's wespeaker model)\n2) It splits the recording into speaker segments using AssemblyAI\n3) It surfaces the most distinctive line from each unnamed speaker \u2014 "Moana is the best movie ever!"\n4) You tap to confirm.
Remember to ask for consent! Ellis prompts you with ways to ask at the start of a recording.
Still not solved: two near-identical voices in the same room (e.g. me and my dad). Curious if anyone found a diarization approach that separates near-identical voices without per-speaker enrollment?
If you have an opinion about the tool, I'm all ears! You can try it for free.","created_at":"2026-08-07T11:34:30Z","created_at_i":1786102470,"objectID":"49208831","parent_id":49208829,"story_id":49208829,"story_title":"Show HN: Cleaner speaker identification for in-person meeting transcripts","story_url":"https://apps.apple.com/us/app/ellis-in-person-ai-notes/id6753657763","updated_at":"2026-08-07T11:39:57Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"bhouston"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"FYI I find that Whisper is far from the best transcription models out there as well. I did a bunch of testing and I found that Assembly AI's transcription models are significant better than Whisper and even ElevenLabs. I guess I now sound like an ad."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Whisper transcribes 70-year-olds more accurately than 20-year-olds"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/Kayvan-Zahiri/asr-age-gap"}},"_tags":["comment","author_bhouston","story_49204388"],"author":"bhouston","comment_text":"FYI I find that Whisper is far from the best transcription models out there as well. I did a bunch of testing and I found that Assembly AI's transcription models are significant better than Whisper and even ElevenLabs. I guess I now sound like an ad.","created_at":"2026-08-07T01:39:00Z","created_at_i":1786066740,"objectID":"49204930","parent_id":49204388,"story_id":49204388,"story_title":"Show HN: Whisper transcribes 70-year-olds more accurately than 20-year-olds","story_url":"https://github.com/Kayvan-Zahiri/asr-age-gap","updated_at":"2026-08-07T01:43:25Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"oletrn"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Startup-minded Senior Full-Stack Developer / Hands-on Technical Product Lead (TypeScript, React, Angular, Next.js, NestJS, Tailwind) with 10+ years of experience in EU/US startups and enterprises across fintech, e-commerce, KYC, and climate tech. Front-end/UI focused but full-stack capable. Built and launched products from 0\u21921, including AI-powered SaaS. Led engineering teams of 5+ engineers. Thrive in async-first remote teams; comfortable working EST hours. Fluent in Japanese, intermediate German.
Location: Prague, Czech Republic (EU)\n Remote: Yes (preferred)\n Relocation: No, open to occasional onsite visits\n Tech: TypeScript, React, Angular, Next.js, RxJS, NgRx, SCSS, Nx, NestJS, NativeScript, Tailwind, CI/CD, REST, MongoDB, PostgreSQL, Jest\n APIs: OpenAI, Stripe, AssemblyAI, DeepL, Twilio, Discord\n Email: oletrn@gmail.com\n Location: Prague, Czech Republic (EU)\n Remote: Yes (preferred)\n Relocation: No, open to occasional onsite visits\n Tech: TypeScript, React, Angular, Next.js, RxJS, NgRx, SCSS, Nx, NestJS, NativeScript, Tailwind, CI/CD, REST, MongoDB, PostgreSQL, Jest\n APIs: OpenAI, Stripe, AssemblyAI, DeepL, Twilio, Discord\n Email: oletrn@gmail.com\nIt's a PWA (no install needed) with a real-time voice pipeline: streaming speech-to-text -> an LLM tutor that keeps an unscripted conversation at your level and corrects you -> low-latency text-to-speech back. The part I care most about isn't the chat \u2014 it's the loop around it: every word you stumble on is auto-captured, then resurfaced later with a Leitner spaced-repetition schedule and silent flashcards, so talking actually turns into retention. It also persists transcripts and seeds each new session with a compact recap of recent ones (a few summaries + the tail of the last conversation, ~1k tokens, prompt-cached), so it continues across sessions instead of resetting \u2014 happy to go into the recall/token-budget tradeoffs.
Stack: Pipecat for the voice pipeline (AssemblyAI STT, an Anthropic model for the tutor with prompt caching, ElevenLabs TTS), FastAPI + Postgres, React PWA, on a single VM because WebRTC media needs UDP. Happy to go deep on the latency tradeoffs and how the SRS interacts with the conversation.
There's a 10-minute trial with no signup wall to start talking. I'd love feedback on the conversation quality and the corrections - and on whether the spaced-repetition loop feels useful or gimmicky."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Langusta \u2013 an AI voice tutor for practicing spoken languages (PWA)"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://langusta.me/"}},"_tags":["comment","author_grajo","story_48552199"],"author":"grajo","children":[48552637],"comment_text":"I built Langusta because I could read Spanish but froze whenever I had to speak it, and booking human tutors meant scheduling + a fair bit of embarrassment.
It's a PWA (no install needed) with a real-time voice pipeline: streaming speech-to-text -> an LLM tutor that keeps an unscripted conversation at your level and corrects you -> low-latency text-to-speech back. The part I care most about isn't the chat \u2014 it's the loop around it: every word you stumble on is auto-captured, then resurfaced later with a Leitner spaced-repetition schedule and silent flashcards, so talking actually turns into retention. It also persists transcripts and seeds each new session with a compact recap of recent ones (a few summaries + the tail of the last conversation, ~1k tokens, prompt-cached), so it continues across sessions instead of resetting \u2014 happy to go into the recall/token-budget tradeoffs.
Stack: Pipecat for the voice pipeline (AssemblyAI STT, an Anthropic model for the tutor with prompt caching, ElevenLabs TTS), FastAPI + Postgres, React PWA, on a single VM because WebRTC media needs UDP. Happy to go deep on the latency tradeoffs and how the SRS interacts with the conversation.
There's a 10-minute trial with no signup wall to start talking. I'd love feedback on the conversation quality and the corrections - and on whether the spaced-repetition loop feels useful or gimmicky.","created_at":"2026-06-16T08:22:54Z","created_at_i":1781598174,"objectID":"48552220","parent_id":48552199,"story_id":48552199,"story_title":"Show HN: Langusta \u2013 an AI voice tutor for practicing spoken languages (PWA)","story_url":"https://langusta.me/","updated_at":"2026-06-17T06:52:53Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"triyambakam"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"My stack has been QuickTime and Assembly AI lol"},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Trace \u2013 Offline Mac meeting transcripts you can flag mid-call"},"story_url":{"matchLevel":"none","matchedWords":[],"value":"https://traceapp.info"}},"_tags":["comment","author_triyambakam","story_48521236"],"author":"triyambakam","comment_text":"My stack has been QuickTime and Assembly AI lol","created_at":"2026-06-15T06:00:01Z","created_at_i":1781503201,"objectID":"48537154","parent_id":48521236,"story_id":48521236,"story_title":"Show HN: Trace \u2013 Offline Mac meeting transcripts you can flag mid-call","story_url":"https://traceapp.info","updated_at":"2026-06-16T04:53:01Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"oletrn"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Startup-minded Senior Full-Stack Developer / Hands-on Technical Product Lead (TypeScript, React, Angular, Next.js, NestJS, Tailwind) with 10+ years of experience in EU/US startups and enterprises across fintech, e-commerce, KYC, and climate tech. Front-end/UI focused but full-stack capable. Built and launched products from 0\u21921, including AI-powered SaaS. Led engineering teams of 5+ engineers. Thrive in async-first remote teams; comfortable working EST hours. Fluent in Japanese, intermediate German.
Location: Prague, Czech Republic (EU)\n Remote: Yes (preferred)\n Relocation: No, open to occasional onsite visits\n Tech: TypeScript, React, Angular, Next.js, RxJS, NgRx, SCSS, Nx, NestJS, NativeScript, Tailwind, CI/CD, REST, MongoDB, PostgreSQL, Jest\n APIs: OpenAI, Stripe, AssemblyAI, DeepL, Twilio, Discord\n Email: oletrn@gmail.com\n Location: Prague, Czech Republic (EU)\n Remote: Yes (preferred)\n Relocation: No, open to occasional onsite visits\n Tech: TypeScript, React, Angular, Next.js, RxJS, NgRx, SCSS, Nx, NestJS, NativeScript, Tailwind, CI/CD, REST, MongoDB, PostgreSQL, Jest\n APIs: OpenAI, Stripe, AssemblyAI, DeepL, Twilio, Discord\n Email: oletrn@gmail.com\nSenior fullstack engineer, 11+ years (mostly Node / TypeScript; 3 years Go), 4+ years as Tech Lead / CTO across remote-distributed teams. Shipped the first alpha of Sophia Forge solo in ~2 months \u2014 a local-first desktop game production app (Electron + NestJS + React) with 5,000+ tests across unit, e2e, and Playwright layers, 30 plugins, a bidirectional Unreal Engine 5 C++ integration, and an MCP-based AI surface. I work AI-native: I direct AI coding agents (Claude Code) on production code, gated by tests, typed contracts, and an LLM-as-judge step. Prior production work includes the UK Parliament real-time captioning platform and multi-provider speech-to-text systems (Twilio, Rev.ai, Deepgram, AssemblyAI). Open to: senior fullstack contract / fractional engagements (immediate availability) or full-time roles where remote-from-Brazil works."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Who wants to be hired? (June 2026)"}},"_tags":["comment","author_sakushimei","story_48357724"],"author":"sakushimei","comment_text":"Location: Caraguatatuba, SP, Brazil (BRT / UTC-3 \u2014 overlaps US East and full EU business hours)\nRemote: Yes\nWilling to relocate: No\nTechnologies: TypeScript, Node.js, Go, NestJS, Next.js, React, Redux, Electron, REST, WebSocket, PostgreSQL, SQLite, MySQL, Redis, AWS (EC2, ALB, RDS), GCP, Docker, Jest, Vitest, Playwright, Claude Code, Model Context Protocol (MCP), OpenAI API, Unreal Engine 5 (+ C++ plugins)\nR\u00e9sum\u00e9/CV: https://drive.google.com/file/d/1tALwOqSIqgmj4ADhvFvgrsq8wkU...\nEmail: jessyka.alycia@gmail.com
Senior fullstack engineer, 11+ years (mostly Node / TypeScript; 3 years Go), 4+ years as Tech Lead / CTO across remote-distributed teams. Shipped the first alpha of Sophia Forge solo in ~2 months \u2014 a local-first desktop game production app (Electron + NestJS + React) with 5,000+ tests across unit, e2e, and Playwright layers, 30 plugins, a bidirectional Unreal Engine 5 C++ integration, and an MCP-based AI surface. I work AI-native: I direct AI coding agents (Claude Code) on production code, gated by tests, typed contracts, and an LLM-as-judge step. Prior production work includes the UK Parliament real-time captioning platform and multi-provider speech-to-text systems (Twilio, Rev.ai, Deepgram, AssemblyAI). Open to: senior fullstack contract / fractional engagements (immediate availability) or full-time roles where remote-from-Brazil works.","created_at":"2026-06-02T01:21:00Z","created_at_i":1780363260,"objectID":"48364739","parent_id":48357724,"story_id":48357724,"story_title":"Ask HN: Who wants to be hired? (June 2026)","updated_at":"2026-06-02T06:06:21Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rpuritty"},"comment_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["assemblyai"],"value":"Hotwash | Founding Engineer (\u2192 CTO for the right person) | REMOTE (US), Boston a plus | Full-time
I'm a solo founder. I built the entire product myself, got 11 fire departments paying with zero churn, and now I need someone to own engineering so I can't be the only one who does. West Point, Ranger Regiment, Yale Law, still the only developer.
What we do: Hotwash is an after-action review platform for fire departments. After a serious call, crews record what happened, we transcribe and structure it, and the hard-won lessons become searchable institutional memory instead of retiring when the veterans who earned them do. Fire is the wedge; the long game is operational learning for any high-reliability organization.
Traction: $40K ARR, 11 paying fire departments, zero churn. Raising a $1M pre-seed.
The role: Employee #1 on engineering. You own the codebase, architecture, and UX end to end, and you lead the rebuild off the current stack. You work directly with me on what gets built and why, including sitting with fire chiefs to figure out what they actually need. For the right person this becomes the CTO seat.
Who I'm looking for: A full-stack engineer who ships simple, usable products and has a real eye for design. Our users are fire crews, not power users, so the bar is "obvious to someone who's never seen it." If you care about how software feels, not just whether it works, this is your seat.
Stack: Today it's FlutterFlow + Firebase, Node, with Groq/AssemblyAI for transcription and Gemini for summarization. The rebuild target is React Native (Expo), Cloudflare Workers, D1, R2. You'd own that migration, so this is real greenfield ownership, not maintenance.
Comp: Founding equity, scaled to scope.
We're pre-raise, so cash is lean right now and grows when we close.
Email hiring@hotwash.com with a resume, link to something you've built, and one line on why this interests you."},"story_title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Who is hiring? (June 2026)"}},"_tags":["comment","author_rpuritty","story_48357725"],"author":"rpuritty","children":[48360282,48361667,48366295,48383986,48536207],"comment_text":"Hotwash | Founding Engineer (\u2192 CTO for the right person) | REMOTE (US), Boston a plus | Full-time
I'm a solo founder. I built the entire product myself, got 11 fire departments paying with zero churn, and now I need someone to own engineering so I can't be the only one who does. West Point, Ranger Regiment, Yale Law, still the only developer.
What we do: Hotwash is an after-action review platform for fire departments. After a serious call, crews record what happened, we transcribe and structure it, and the hard-won lessons become searchable institutional memory instead of retiring when the veterans who earned them do. Fire is the wedge; the long game is operational learning for any high-reliability organization.
Traction: $40K ARR, 11 paying fire departments, zero churn. Raising a $1M pre-seed.
The role: Employee #1 on engineering. You own the codebase, architecture, and UX end to end, and you lead the rebuild off the current stack. You work directly with me on what gets built and why, including sitting with fire chiefs to figure out what they actually need. For the right person this becomes the CTO seat.
Who I'm looking for: A full-stack engineer who ships simple, usable products and has a real eye for design. Our users are fire crews, not power users, so the bar is "obvious to someone who's never seen it." If you care about how software feels, not just whether it works, this is your seat.
Stack: Today it's FlutterFlow + Firebase, Node, with Groq/AssemblyAI for transcription and Gemini for summarization. The rebuild target is React Native (Expo), Cloudflare Workers, D1, R2. You'd own that migration, so this is real greenfield ownership, not maintenance.
Comp: Founding equity, scaled to scope.
We're pre-raise, so cash is lean right now and grows when we close.
Email hiring@hotwash.com with a resume, link to something you've built, and one line on why this interests you.","created_at":"2026-06-01T15:39:03Z","created_at_i":1780328343,"objectID":"48358328","parent_id":48357725,"story_id":48357725,"story_title":"Ask HN: Who is hiring? (June 2026)","updated_at":"2026-09-29T19:06:09Z"}],"hitsPerPage":20,"nbHits":254,"nbPages":13,"page":0,"params":"query=AssemblyAI&tags=comment&advancedSyntax=true&analyticsTags=backend","processingTimeMS":28,"processingTimingsMS":{"_request":{"queue":2,"roundTrip":14},"afterFetch":{"format":{"highlighting":1,"total":2},"merge":{"mergeLoop":{"prepareNextHit":2,"total":2},"total":2},"total":2},"fetch":{"query":14,"scanning":10,"total":25},"total":28},"query":"AssemblyAI","serverTimeMS":32}