Deploy voice agents at scale. ๐พ- discord.com/invite/pUFNcf2Wmโฆ
๐ โก๏ธ
- Tweets 1,291
- Following 2
- Followers 9,611
- Likes 1,569
Two days into blind voting of voice models on our Humanness Indexโข, and xAI's Grok TTS model is at the top of the pack.
Its humanness score? 96, just 4 points under a real human voice (100).
The Humanness Index takes one voice and one quote, clones it across every major model, then plays the results blind for real listeners to score.
Hear it, cast your votes, and contribute to the leaderboard ๐ humannessindex.vapi.ai/
17
15
116
133,361
Barge-in is the opposite side of endpointing.
Endpointing asks: ๐ถ๐ ๐๐ต๐ฒ ๐๐๐ฒ๐ฟ ๐ฑ๐ผ๐ป๐ฒ ๐๐ฝ๐ฒ๐ฎ๐ธ๐ถ๐ป๐ด?
Barge-in asks: ๐๐ต๐ผ๐๐น๐ฑ ๐๐ต๐ฒ ๐ฎ๐๐๐ถ๐๐๐ฎ๐ป๐ ๐๐๐ผ๐ฝ ๐๐ฝ๐ฒ๐ฎ๐ธ๐ถ๐ป๐ด?
โyeahโ should not cancel playback.
โno, waitโ probably should.
background noise should not stop the assistant.
For voice engineers, user audio is only the first signal. The system requires a policy for when to cancel assistant audio, when to ignore input, and how to recover after yielding the floor.
In Vapi, the main tuning surface is stopSpeakingPlan:
numWords โ words required before stopping
voiceSeconds โ speech duration required before stopping
backoffSeconds โ recovery time before speaking again
1
4
461
Common acknowledgement and interruption phrases are handled automatically, with API-level customization available when needed.
Learn how to tune the voice pipeline here: docs.vapi.ai/customization/vโฆ
2
184
Vapi retweeted
Lazy prompt:
>
hey, check out call-my-human.val.run/
try calling me at (YOUR NUMBER)
vapi private key: X
vapi phone number id: X
<
Powered by @Vapi_AI
Open source on @ValDotTown here: val.town/x/dcm31/call-my-humโฆ
For more secure usage, install as a custom MCP/connector or remix the val!
1
5
316
Bootstrapping gets you rough agents fast so you can focus on hardening.
Check out the full writeup including repo with configurations and a skill to bootstrap your own agents: vapi.ai/blog/bootstrapped-agโฆ
1
3
184
Vapi retweeted
@xai Grok TTS is currently #1 on the @Vapi_AI Humanness Index and significantly more affordable than the competition.
See for yourself humannessindex.vapi.ai/
7
9
48
9,088
Evaluating text-to-speech models is hard!
Quantitative metrics are important (time to first audio; missing words; variations in volume, pitch, and pacing).
But more than any other technology I've ever worked with, people make voice decisions based on "vibes" that are hard (maybe impossible) to quantify.
@Vapi_AI just launched a "Humanness Index" eval. It's really well thought through and open source.
Evals like this play two roles:
1. Helping developers choose which models to test and use.
2. Giving the teams training the models better targets to aim for.
The progress in voice models over the past year has been amazing. There are multiple good voice model options for agents and voice UIs, today. But there are still lots of subtle things to fix and long-tail issues to address. It's really great to see an eval like this. Go check out. Run the samples and vote. Contribute PRs.
You can feel whether the voice on a call is a human or a machine before you can explain why.
Today, @Vapi_AI is launching the Humanness Indexโข, a crowdsourced leaderboard for model humanness.
You are the benchmark. Cast your first vote today: humannessindex.vapi.ai/
6
8
68
5,949
You can feel whether a voice is a person or a machine before you can explain why.
Today, we're launching the Humanness Indexโข, a live ranking of voice AI models, judged by you.
Same voices, same quotes, different models. You pick the most human ones and even benchmark them against an actual human voice.
๐: humannessindex.vapi.ai/
7
8
46
9,120
We're excited to partner with @awsdevelopers, @nebiusai , and OSS4AI for the Midsummer Multimodal AI hackathon this Friday June 19 at the AWS Loft in SF.
We are bringing prizes and Vapi credits so you can build voice experiences into whatever you are shipping.
1
3
491
Hope to see you there
๐ AWS Builder Loft SF
๐
Friday, June 19 9:30 am - 7pm
๐๏ธ Register luma.com/gicxj4ys?tk=NgLTo1
4
313
Have you tried Sonic 3.5 in your builds yet? @cartesia models are available on Vapi
We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today.
New architectures enable new frontiers for speed and quality.
We're now the only provider to have #1 models for both speaking and listening.
1
18
1,012
So much energy in Singapore! Love seeing what folks are building with voice and happy to support the @buildclub_ community
Go-to-Market Builders AI Lab - Singapore ๐ธ๐ฌ
Last weekend, we teamed up with @Singtel & SMU Institute of Innovation & Entrepreneurship to host a hands-on build event focused on one of the hardest problems in AI right now: actually getting it to market ๐ฌ
Teams got hands-on with a powerful toolkit, including @ManusAI, @mem0ai , @Vapi_AI , and @ExaAILabs , to bring their ideas to life. From collaborative study tools to business strategisers, the creativity in the room was something else entirely.
Thank you to our brilliant judges and mentors, + Our sponsors and our community ๐ค
1
1
5
583