AI Company Sesame has released the basic model that dominates Maya, the impressive realistic voice assistant.
The model, which is 1 billion in size parameters (“parameters” referring to individual elements of the model), is under license Apache 2.0, which means that it can be used commercially with few restrictions. Named CSM-1B, the model creates “RVQ audio codes” from text and audio inputs, according to SESAME’s description in AI Dev Platform Hoging Face.
RVQ refers to the “residual quantum vector”, a sound coding technique in distinct brands called codes. RVQ is used In many recent Ai Audio technologiesIncluding Google’s Soundstream and Meta’s Encodec.
CSM-1B uses a model from Meta’s Llama family as its backbone in combination with an “decoder” headset. A delicate variant of CSM Powers Maya, Sesame says.
“The model open-extension here is a generation of base,” writes Sesame on CSM-1B Hug and Github repository. “He is able to produce a variety of voices but has not been perfected in any particular voice […] The model has some capacity for non -English languages due to data contamination in training data, but it will probably not do well. ”
It is not clear what data the sesame is used for CSM-1B training. The company didn’t say.
It is worth noting that the model has no real safeguards to speak. Sesame has a price system and simply urges developers and users not to use the model to imitate a person’s voice without their consent, to create misleading content such as false news or to participate in “harmful” or “malicious” activities.
I tried the demo In hugging face and cloning of my voice it needed less than a minute. From there, it was easy to create a speech in the desire of my heart, including controversial issues such as elections and Russian propaganda.
Consumer reports recently warned that many popular voice cloning tools You don’t have “important” safeguards to prevent fraud or abuse.
SESAME, co-founded by Oculus Brendan Iribe’s co-creator, became viral at the end of February for the technology assistant, approaching the clearance of Uncanny Valley. The other assistant of Maya and Sesame, Miles, breathe and speak with disorders and can be interrupted while talking, as does Openai’s voice function.
SESAME has set an unprecedented amount of capital by Andreessen Horowitz, Spark Capital and Matrix Partners. In addition to building Voice Assistant Tech, the company says they are original AI glasses “designed to be worn all day” equipped with its custom models.