hi guys have a client in Indonesia and Philippines who loved our English voice AI demo, but now wants one in Bahasa and Philippine English/Taglish mixed language, code switching and all
anyone here have real experience deploying voice agents in SEA, specifically TTS that handles code switching well?
what's actually held up in production what sounds good in a demo
found this benchmark from kwindla the co founder of pipecat
repo https://github.com/kwindla/aiewf-eval
ultravox tops the s2s leaderboard beats gpt realtime 2
is it go to for s2s now??
I’m considering a tts for one of the European languages and my current provider cartesia isn’t really it
I see inworld they have their own platform too but everyone does these days but anyways any feedback is appreciated
we are still using gemini flash live 2.5 for our biggest client can’t move to 3.1
is there a workaround, we considering switching I’ve already wrote in the community about it google docs don’t have no plans to fix it
has anyone tried with pipecat voximplant agora does the issue persist
I saw they ran some test comparing the voice agent platforms vapi livekit pipecat retell synth flow and retell
And put retell at first
Meanwhile Retell is an official partner of theirs and so are other top names on the list while they put none partners below
The list is biased as usual, you shouldn’t trust it
With all the hype around the new models Kimi and the letter to the US government I’m wondering does anyone actually use open source llms in the voice AI industry
I’ve heard of couple of providers of inference together AI baseten and others but please enlighten me if you had any experience with them
I’ve only tried open AI Gemini and grok in production
I ran a sales call for a small client that was using 3cx with human agents for customer support
Anyways the voice AI demo for cx went good he was all happy but I left the price at the end and he thought it wasn’t worth it
he came back couple of days later asking me if we could implement basic ivr with 3cx integration for his team instead
We agreed and the implementation took us couple of hours and we closed the deal
Some people still not ready I guess for AI huh
I build voice agents in romania I don’t have a lot of clients
Im okay with my current stack (it is managed)
I was thinking moving with orchestration or else to open source idk if it really matters
Will it change anything?
I was demoing the product to the client
Ran a call then he asked me to call him using my agent, then he said he didn’t like the latency and asked me if it can be negative
I told him that latency can’t be negative and he just started arguing and told me if he can get into my platform and fix the prompt himself
Anyways, I didn’t close the deal, how do yall go on with this?
I am building a platform I need open source alternative to twilio
Saw this on linkedin and got interested what do these platforms actually do
Isn’t checking logs not enough to see what’s wrong
If anyone had feedback with them I’d appreciate to hear, but it is a marketing move probably since they only put themselves (ig it works)