Copywriter, technical writer, translator (FR>EN, ES>EN, IT>EN), journalist

Sound principles for gathering evidence

They’re not just for making calls anymore. Today’s smartphones create video and still images, and record audio too. Aside from the calling, much the same can be said of tablets.

These two types of multimedia enter businesses every day in the hands of employees who may prefer to use their own gadgets instead of company-issued smartphones.

“In the old days, IT administrators dictated what technology people used,” says Kevin Lo, managing director for Froese Forensic Limited., a financial investigation firm based in Toronto. “Now, they get pushback from people who want to use other things.”

And that means more places for potentially probative multimedia files to reside, outside company-sanctioned content management systems.

Granted, the business reasons for allowing this trend can rival any reasons for disallowing it. “In meetings, I now type notes on my iPad and I can record what’s said at the same time,” Lo says.

“The technology is so casual now. You used to have to open a laptop and start a couple of programs.”

It all goes to the Internet and gets synchronized to my other devices.”

While this setup enhances productivity, it also can cause issues related to confidentiality, privacy and ownership of data that legal professionals are now coming to understand.

Bring your own device

For instance, many companies buy computers and phones for their employees, and all parties understand that the owner of the devices also owns the data on them. However, some companies subsidize employees who purchase devices, understanding they’ll be used for both business and personal reasons. The mixture of personal and business records on these devices makes for a tangled ownership web. an “If the owner of a smartphone takes photos of his girlfriend, who owns those pictures?” Lo asks. He adds that the answer isn’t clear in every case.

But companies can take steps to minimize their exposure to liability. Security and file management policies can now include:

  • forbidding recordings at meetings without the express consent of all attendees; and
  • demanding that employees and contractors leave personal devices, especially mobile phones, in “clean rooms” before they enter areas that house sensitive intellectual property.

Records management policy

Advise clients to regularly purge records according to their records retention policy. Such systems can be set to overwrite sections of storage media that previously contained information. (While tech wizards have ways of retrieving purged data, their services can be expensive. Generally, courts accept that documents are gone if deleted according to a documented, legitimate data-handling policy.)

“You can minimize the cost of discovery if you adhere to your deletion schedule,” says Susan Nickle, co-founder of Wortzman Nickle Professional Corp.

Handling audio files

Software such as Adobe Audition can create a “noise print” and remove a segment of sound from an audio recording. “It’s like having a conversation with street noise, and closing the door to shut out the street noise,” explains Colin Smith, senior solutions architect for Adobe Systems Inc. He adds that the effect is non-destructive and easy to undo.

Similar tools let you “clarify” photos and video as well. That said, companies such as Nexidia want to work with original files only, warts and all.

“[Cleaning up the file] makes it easier on your ears but it also degrades the underlying original content,” says Jeff Schlueter, vice-president of legal markets for Nexidia Inc. “[The process] gets rid of key digital information that lets us get good accuracy using our indexing tool.”

Phonetic audio search

In some industries, such as securities or energy trading, all messages must be recorded by law. Add to that the files created in the normal course of business and substantial amounts of communications pile up.

Should discovery become necessary, most people involved know how to handle text-based material. Audio recordings, though, have been more challenging to examine.

The obvious method is to assign a bunch of people to listen to different segments of audio evidence, which can include phone conversations, wiretaps, voicemail and even specialty platforms such as Skype.

People might detect telling voice inflections that computers would not. But discovery can only proceed at the pace of a listener’s comprehension and ability to transcribe, which makes the process very time-consuming and expensive.

Speech-to-text technology promises to make audio discovery more like discovery of text documents. However, current speech-to-text tools aim for written accuracy, a time-consuming challenge when several voices on a recording speak with different accents.

The current “best-practice” technology foregoes the goal of written accuracy. Instead, it turns audio recordings of conversations into phonetic representations, or phonemes. It also turns user queries into phonemes. With both queries and recordings converted into these phonetic lowest common denominators, the technology then searches the recordings for matching phonemes. “We create a timeline and phonetic representation of the content, to make it searchable,” says Jeff Schlueter, vice-president of legal markets for audio and video search service provider Nexidia Inc. “And we bring in metadata as well.”

He claims Nexidia’s hosted model of “discovery on demand” enables several people to search the audio at the same time. “It works like any other web-based review platform.”

Phonetic search remains imperfect: It returns some false positives among the more highly relevant (and ranked) positives, while leaving some actual posi- tives among lower-ranked finds— somewhat like a Google search which returns one or two irrel- evant results on the first page and one or two relevant results on the tenth page.

But Nexidia claims much quicker, cheaper and more accurate conversion of audio files to searchable content than either human-powered discovery or speech-to-text can achieve.

It included a sample audio discovery request in its white paper “Audio Discovery Guide: Introduction to Requesting and Collecting Audio Data” that people can build on to create their own audio requests.

Model discovery request

Consider using these abbreviated excerpts from Nexidia’s model in your own production requests; many of them apply to photos and video as well as audio (The full white paper can be downloaded from Nexidia.com):

1. Describe any systems used to capture or record sound recordings.

2. Describe connections to any network.

3. Describe any systems used to store or archive sound recordings.

4. Describe software used to record, archive, back up, extract, review, or otherwise process sound recordings.

5. Identify any location, including any local or network drive or other storage device where relevant recordings are stored.

6. Provide a copy of the respondent’s record-retention policies for sound recordings.

7. Describe the software configurations, methods, policies, or business rules applied to the audio recording system.

8. Identify the person or persons most knowledgeable about the administration of the audio recording and storage system as well as the custodian(s) of the sound recordings.

9. Identify all persons who have been granted electronic access rights to sound recordings in either the recording system or storage system.

10. State the amount of capacity currently used to store sound recordings in the recording system as well as the storage system.

This article originally published in The Lawyers Weekly. For a PDF of this article, click here.