Cepstral Opens Protocol Library to Spur Development

Cepstral, a speech technology company, has decided to open its Media Resource Control Protocol (MRCP) library to the developer community.

The move is designed to free telephony application developers from the protocol grindstone and lead to greater inroads in software innovation, Cepstral’s CEO said.

The mission is to spur the development of telephony applications that interact with text-to-speech (TTS) and automatic speech recognition (ASR).

Less Drudgery, More Innovation

The announcement is Cepstral’s effort to free up outside developers’ time and labor. In offering OpenMRCP to the community, Cepstral foresees less drudgery and more software innovation for the outside developer community.

MRCP is a communications protocol that allows telephony applications to communicate with speech resources. “Think TCP/IP (transmission control protocol/Internet protocol),” Cepstral CEO Craig Campbell told LinuxInsider. Whereas TCP/IP is an enabler for browsers to talk to Web sites, MRCP is a protocol for telephone exchanges to communicate with ASR and TTS engines.

Sludge Barrier

What kind of drudgery? Without it, each company’s developers would need to reinvent the protocol layer, said Campbell. That requirement for each company to reinvent the layer was inefficient. “Cepstral hopes OpenMRCP will accelerate the development of speech applications,” he said.

Cepstral knew the gruntwork firsthand.

“We know the pain of not having it ourselves, which convinced us: We shouldn’t be working on protocols, we should be working on our own speech technologies,” Campbell said. That was the deciding factor.

“We developed an MRCP solution for our own use, to serve our own interests, and now we are releasing it for others,” Campbell said.

Speech Pipeline

Call it altruism, call it open source spiritualism, but also call it good corporate strategy. In removing the barriers to entry with OpenMRCP, Cepstral is moving closer to the front lines of marketplace recognition.

“We want to be the text-to-speech people,” Campbell said. “We’re growing our presence in the telephony area.” By open sourcing its MRCP library, Cepstral is moving that much closer to telephony application developers.

Cepstral is building natural-sounding voices for handheld, desktop, and server applications. Industry observers agree that the market potential for the spoken delivery of information is limited only by the imagination.

“It’s not that speech technology is going to replace text on the screen. It’s a multimodal world,” Campbell said.

Leave a Comment

Please sign in to post or reply to a comment. New users create a free account.

LinuxInsider Channels