The field has plenty of competition. Another entrant is Media Audit, a unit of Houston's International Demographics Inc., which has collaborated with Paris-based Ipsos SA on a cellphone-based measurement system being tested by such companies as radio-station owner Clear Channel Communications Inc.

Some companies argue that cellphones could lead to distorted research. Survey participants, for example, could change how often they carry or converse on phones or download content to them.

Arbitron Inc. instead proposes a special-purpose gadget called the portable People Meter, which it has been testing in Houston. GfK AG's Mediamark Research Inc. also is developing a pager-size media-measurement device.

Radio and TV exposure has long been measured, respectively, by Arbitron and VNU NV's Nielsen Media Research. Methods include pen-and-paper usage logs filled out by selected panels of consumers, as well as devices that passively measure media usage in the home.

But consumers' media exposure has taken on many new forms, including delayed TV broadcasts with digital video recorders, Internet video, videogames and songs or movies that are downloaded to Apple Computer Inc.'s iPod or other portable devices. The new measuring services aim to better follow consumers as they move among the media types.

IMMI Chief Executive Tom Zito previously led videogame company Digital Pictures and Garageband.com, a Web music start-up. A partner at both ventures was Amanda Welsh, now an IMMI senior vice president. Al Alcorn, IMMI's chief technology officer, co-founded videogame pioneer Atari Corp.

The entrepreneurs, backed by $14 million in venture capital, plan to give survey participants cellphones that can take reliable sound samples, Mr. Zito said. Those snippets -- taken every 30 seconds and altered mathematically so any conversation is made unintelligible -- are transmitted continuously to IMMI. Sounds from headphone devices such as iPods can be transmitted to the cellphones with a wireless accessory. IMMI has been building a database of sound signatures, with help from customers testing the company's services as well as with CD content it has licensed.

Arbitron and some other companies prefer an approach that uses inaudible identifying code, sometimes called a watermark, that can be inserted into broadcast programming. Arbitron spokesman Thom Mocarsky argues that matching audio samples doesn't work, say, if two radio stations are airing the same syndicated program.

But encoding a watermark also has downsides, Mr. Zito says, including requiring time and effort on the part of content providers -- and embedding such identifiers isn't possible on DVDs and CDs that have already been sold.

Dave Harkness, a senior vice president of strategy and business development at Nielsen, said both watermarks and sampling may be needed to provide the most complete exposure ratings. But sampling alone may still provide valuable information to advertisers and media companies, he said.

IMMI's Mr. Zito says there is evidence that sampling provides useful clues about ad effectiveness. Some respondents in the Sacramento test who were exposed to a particular trailer for the movie "King Kong" went to see the film more than twice as often as those who saw an alternate trailer for the movie, IMMI's data show.

With conventional media ratings, "we don't have the opportunity to take a look at cause and effect," said Artie Bulgrin, senior vice president of research and sales development at ESPN, which has been evaluating IMMI's service. "What we see so far is very interesting and very compelling."

Write to Don Clark at don.clark@wsj.com¹