Topics
late
AI
Amazon
Image Credits:Bryce Durbin
Apps
Biotech & Health
Climate
Image Credits:Bryce Durbin
Cloud Computing
commercialism
Crypto
PlayAI’s agent feature, which builds automation tools around the company’s text-to-speech engine.Image Credits:PlayAI
Enterprise
EVs
Fintech
PlayAI’s PlayDialog model can generate two-day, “duplex” conversations that sound relatively natural.Image Credits:PlayAI
Fundraising
widget
bet on
Government & Policy
Hardware
layoff
Media & Entertainment
Meta
Microsoft
privateness
Robotics
security system
Social
Space
Startups
TikTok
Transportation
speculation
More from TechCrunch
event
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
Back in 2016 , Hammad Syed and Mahmoud Felfel , an x - WhatsApp engineer , thought it ’d be neat to build a text - to - speech Chrome extension service for average articles . The extension phone , which could read any Medium story loudly , was featured on Product Hunt . A yr by and by , it spawned an entire business organization .
“ We saw a grown chance in assist individuals and organizations create realistic audio depicted object for their applications , ” Syed told TechCrunch . “ Without the need to build up their own manikin , they could deploy human - lineament speech communication experiences faster than ever before . ”
Syed and Felfel ’s caller , PlayAI(formerly PlayHT ) , flip itself as the “ voice port of AI . ” Customers can choose from a number of predefined voices , or clone a vocalism , and use PlayAI ’s API to integrate text - to - language into their apps .
Toggles permit substance abuser to adjust the intonation , cadence , and tenor of articulation .
PlayAI also offers a “ playground ” where users can upload a file to father a read - aloud version and a splasher for creating more - dressed audio narrations and part - overs . Recently , the company got into the “ AI agent ” game with tools that can be used to automate task such as answering customer scream at a business .
One of PlayAI ’s more interesting experiments is PlayNote , which transforms PDFs , videos , photos , song , and other files into podcast - style shows , read - aloud summaries , one - on - one debates , and even children ’s story . Like Google’sNotebookLM , PlayNote get a script from an uploaded Indian file or uniform resource locator and give it to a collection of AI mannikin , which together craft the ruined product .
I gave it a whirl , and the answer were n’t half bad . PlayNote ’s “ podcast ” setting produces cartridge holder more or less on equality with NotebookLM ’s in term of lineament , and the cock ’s power to ingest photos and television makes for some fascinating creations . give a picture of a chicken mole dish antenna I had late , PlayNote wrote a five - minute of arc podcast script about it . rightfully , we are living in the time to come .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
grant , the tool , like all AI instrument , generates odd artifacts andhallucinationsfrom prison term to time . And while PlayNote will do its unspoilt to adapt a file to the format you ’ve chosen , do n’t wait , say , a wry sound filing to make for the best rootage fabric . See : theMusk vs. OpenAI lawsuitframed as a bedtime story :
PlayNote ’s podcast format is made possible by PlayAI ’s latest model , PlayDialog , which Syed enunciate can use the “ context and history ” of a conversation to render speech that chew over the conversation stream . “ Using a conversation ’s historical context to control prosody , emotion , and pacing , PlayDialog deliver conversation with lifelike deliverance and appropriate feeling , ” he continued .
PlayAI , which is near competition with ElevenLabs , has beencriticizedin the past tense for its laissez - faire glide slope to safety . The company ’s voice cloning tool requires that users go over a boxwood indicating that they “ have all the necessary rights or consent ” to clone a voice — but there is n’t any enforcement mechanism . I had no trouble make a clone of Kamala Harris ’ voice from a recording .
That ’s concerning considering thepotential for scamsanddeepfakes .
PlayAI also claim that it automatically notice and blocks “ sexual , offensive , racist , or threaten content . ” But that was n’t the casing in my examination . I used the Harris clone to mother speech I frankly ca n’t plant here and never once saw a warning subject matter .
Meanwhile , PlayNote ’s community portal , which is filled with publically mother content , has files withexplicit titleslike “ Woman execute Oral Sex . ”
Syed tells me that PlayAI responds to reports of voices clone without consent , like this one , by jam the user responsible and removing the cloned voice directly . He also makes the case that PlayAI ’s high - fidelity voice clones , which require 20 minutes of voice samples , are priced higher ( $ 49 per calendar month billed annually or $ 99 per calendar month ) than most defrauder are unforced to pay .
“ PlayAI has several ethical precaution in spot , ” Syed aver . “ We ’ve implemented robust mechanisms to identify whether a voice was synthesize using our technology , for exemplar . If any misuse is reported , we quickly swear the descent of the content and take decisive actions to rectify the situation and prevent further ethical violations . ”
I ’d for sure trust that ’s the fount — and that PlayAI proceed away frommarketing hunting expedition featuring dead tech fame . If PlayAI ’s relief is n’t robust , it could confront effectual challenges inTennessee , which has a law on the book keep program from host AI to make unauthorised recordings of a person ’s representative .
PlayAI ’s approach to school its voice - cloning AI is also a bit murky . The company wo n’t reveal where it source the data for its modeling , ostensibly for free-enterprise rationality .
“ PlayAI uses mostly open datasets , [ as well as licensed data point ] and proprietary datasets that are built in - house , ” Syed said . “ We do n’t utilise user data from the products in training , or creators to train models . Our models are trained on trillion of hours of substantial - biography human delivery , delivering voices in male and female sexuality across multiple languages and accents . ”
Most AI modeling are trained on public vane data — some of which may be copyrighted or under a restrictive permission . Many AI trafficker argue that thefair - usedoctrine shields them from copyright claims . But that has n’t stopped data ownersfromfiling grade legal action lawsuits alleging that vendors used their information sans permission .
PlayAI has n’t been sued . However , its term of servicesuggestit wo n’t go to bat for drug user if they observe themselves under legal threat .
Voice cloning political program like PlayAI face criticism from actors who fear that voice work will eventually be supplant by AI - generate vocals and that actors will have small control over how their digital doubles are used .
The Hollywood worker ’ union SAG - AFTRA has struck peck with some startups , including online gift market Narrativ and Replica Studios , for what it describes as “ fair ” and “ honorable ” voice cloning arrangements . But even these tie - ups have arrive underintense scrutiny , including from SAG - AFTRA ’s own members .
In California , jurisprudence require company relying on a performer ’s digital replica ( e.g. , cloned vocalism ) give a description of the replica ’s mean usance and talk terms with the performer ’s legal counsel . They also require that amusement employers gain the consent of a deceased performer ’s estate before using a digital ringer of that person .
Syed says that PlayAI “ guarantees ” that every voice clone mother through its platform is exclusive to the Divine . “ This exclusivity is vital for protect the creative rightfield of users , ” he added .
The increase legal burden is one headwind for PlayAI . Another is the rival . Papercup , Deepdub , Acapela , Respeecher , andVoice.ai , as well as Big Tech incumbents Amazon , Microsoft , and Google , offer AI dubbing and voice - cloning prick . The said ElevenLabs , one of the high - profile voice - cloning seller , is said to beraisingnew funds at a rating over $ 3 billion .
PlayAI is n’t struggling to find investor , though . This month , the Y Combinator - backed company close a $ 21 million seed round co - lead by 500 Startups and Kindred Ventures with participation from Race Capital , 500 Global , and Soma Capital .
“ The new uppercase will be used to enthrone in our generative AI voice models and part federal agent platform , and to reduce the time for businesses to build human - quality speech experiences , ” Syed enunciate , adding that PlayAI plans to extend its 40 - somebody workforce .