E-Discovery Update - by Fios Inc.: Assessing the Importance of Voice Mail In Discovery

By Conrad J. Jacoby, Published on May 20, 2006

Discovery of voice mail messages in civil litigation has been increasing over the past few years. Once generally ignored-or available only at significant expense-voice mail messages are recognized today as interesting sources of unfiltered information that may not be reflected elsewhere in written records. Though not all litigation matters or internal investigations would gather substantive, unique information by reviewing voice mail messages, at the very least, these electronic documents are now an information source that should be considered in developing any discovery plan.

A major driving force behind voice mail discovery is the increasing ease with which this information can be harvested. In the past, the complexities of voice mail systems meant that it was challenging, even for a qualified forensic examiner, to identify and extract voice mail messages so that they could be reviewed for substantive content. As a result, discovery requests involving voice mail could often be successfully challenged on grounds of the significant burden they imposed on the responding party. In contrast, today's computerized phone systems typically store voice mail messages as computer sound files (e.g., .WAV files) that, in theory, can be preserved and copied much like other digital data files.

Telephone equipment manufacturers are concerned about hackers and information security, and they are reluctant to release full technical specifications for accessing their proprietary voice mail systems outside the professional telephony community. However, many of today's systems are based on some variant of UNIX or another standard computer operating system. That means that computer forensics professionals can often harvest raw data files from a modern voice mail system, even if they don't necessarily understand the nuances of the application software that creates voice mail message files. However, specialized knowledge may be required if the voice mail system routinely encrypts stored voice mail messages. In such systems, raw data files can be harvested or preserved, but they will not be particularly useful unless they are decrypted.

Voice mail messages can also be collected relatively easily from organizations that have integrated their voice mail and e-mail systems. These systems forward voice mail messages as they are left to e-mail servers as attachments to e-mail messages. Depending on the type of integration, such automatically-generated e-mail messages can include such information as the incoming phone number, the date and time of the call, the length of the message, and whether the message was flagged by the caller as high-priority. Voice mail messages that are passed to e-mail systems in this way are already collected today in sweeps of e-mail systems, though the audio file attachments are not routinely processed further in many cases.

Harvesting computer files is only the first step in analyzing voice mail, however. Unlike text-based documents that are stored on corporate data networks, voice mail messages are stored as audio files. These sound files have to be played-and heard-one by one in order to understand their potential relevance and significance. Some litigation teams have assembled document review teams to review voice mail messages, but because most people read far more quickly than we speak or process aural information, listening to voice mail messages takes substantially longer than reviewing the same content if it were written out.

Recognizing the need for greater efficiency, some litigation teams have taken the additional step of sending out harvested voice mail messages for transcription by outside vendors. These vendors use inexpensive hourly workers to listen to these messages, and return written transcripts that can be searched much more quickly by the review team. This strategy is promising, but it also presents some shortcomings. First, voice mail messages can be difficult to understand. Regional accents, elevated emotions (e.g., shouting, crying), and foreign languages can all make it difficult for an outsider to fully capture a message. In addition, voice mail messages may use proper names, shorthand expressions, or other statements that are difficult to understand and accurately capture. Since these names and expressions may be the important part of a message, poor transcription of this content may limit the overall value of transcribed messages.

Technology offers a final promising (and still developing) alternative to human review of voice mail messages. Several companies are pioneering the application of voice recognition technology to undifferentiated accumulations of voice mail. Normally, voice recognition software must be trained to a specific human voice (some would argue that the humans actually train themselves to match the software) in order to accurately capture its words. Alternatively, as seen in many voice-driven customer support lines, computer software can be trained to recognize a limited vocabulary (e.g., "yes," "no," "help") that is spoken by a wide variety of voices. Automated voice recognition software uses a combination of algorithms and raw computing power to generate increasingly accurate transcriptions of voice mail messages. While this technique is not consistently reliable in all cases, the automated process provides initial insight into the voice mail message population at an attractive price.

Is Voice Mail Appropriate Discovery in Your Matter?

Though voice mail is much more accessible today than it was only a few years ago, working with these electronic files remains much more cumbersome than working with other digital information like e-mail messages and Microsoft Office documents. As a result, producing parties may have strong, valid reasons to resist producing this material. Conversely, receiving parties may find that they must spend substantial amounts of money to transcribe a large number of voice mail messages in order to find-or eliminate the potential for-one or two messages of interest. Several basic questions will help a litigation team decide whether the potential value of this material is likely to be outweighed by the burden of collecting it and making it accessible for review.

1. How Do People Communicate Within the Organization?

Different corporate cultures develop clear preferences in how employees communicate. In some businesses, wide availability of Blackberry, Treo, and other smartphones has encouraged employees to correspond via e-mail or text message-especially if they think their exact words may be important in the future. Elsewhere, a company with a robust "open door" policy may encourage employees to communicate face to face, rather than through the faceless medium of phone or e-mail. And of course, at some places, voice mail may be the only way that certain employees can connect.

Some organizations, in an attempt to reduce reliance on voice mail, have propounded formal policies that discourage employees from leaving anything other than their name, phone number, and time of day in a voice mail message. While such limiting policies may be honored in the breach by employees, these companies' affirmative measures are intended to reduce the need for them to review voice mail messages for potential relevance to any litigation matters. Evidence of these policies and ways in which companies enforce these policies can be a powerful demonstration of why voice mail discovery is unlikely to obtain substantive evidence.
Attorneys trying to understand the potential value-or lack of value-of voice mail messages need to investigate how key players in their matter have communicated. Interviews and depositions are one way of asking key individuals-and their colleagues and assistants-about their personal communication practices. Another avenue of inquiry is to seek information about how many voice mail messages are left for key players on a daily or weekly basis. A combination of these two strategies should quickly reveal the extent to which key players in a legal dispute have relied upon voice mail to communicate.

2. How Specific Are The Discovery Requests?

Because collecting and processing voice mail messages remains more challenging and expensive than many other "active data" stored by an organization, both requesting and producing parties should understand the burden of complying with the voice mail discovery that is being sought. This is a highly fact-specific analysis. For example, while a discovery request for "all communications between persons Smith, Jones, and Campbell" may seem consistent with general discovery practice, it may be extraordinarily difficult to apply this request to voice mail messages. Voice mail systems are designed for security, not for ease of raw information retrieval, and systems may disguise or encrypt metadata such as incoming phone number available, making it difficult or impossible to filter messages in their native format. As a result, a producing party may have to harvest a very large universe of voice mail messages from which to glean the few messages that might meet discovery criteria. In more than a few cases, the substantive value of this information could be significantly outweighed by the expense and burden of obtaining it.

In addition to general burden, requesting parties should also consider the breadth of their voice mail discovery requests. Unlike e-mail messages and other text-based digital information, voice mail messages cannot be filtered by keyword or other content until fairly late in the processing cycle, after considerable work and expense. Thus, even basic relevancy analysis could be an expensive proposition. Requests for "all communications pertaining to X, Y, or Z or containing the keywords "guilt," "fraud," and "lies"" could require preserving and processing an entire organization's voice mail files so that key word filtering can be applied. Courts have been sympathetic to the burden and unclear benefit of such discovery requests when they understand that an enormous volume of clearly irrelevant material must be processed-at significant cost-in order to confirm or eliminate the possibility of relevant voice mail messages.

Requesting specific voice mail boxes has often been upheld by courts as targeted discovery of reasonable scope. In addition, depending on the procedures approved by the court for harvesting the voice mail messages, several "low-tech" methods can be applied to collect the relatively limited number of voice mail messages found in a discrete number of voice mail accounts.

3. How Accessible Are The Voice Mail Messages?

As noted previously, a significant test for the discoverability of voice mail messages is the amount of effort it will take to harvest this information. Depending on how voice mail messages are handled at an organization, it may be relatively easy-or profoundly difficult-to collect this information.
Although it may not be considered forensically sound by all courts, the "lowest impact" method for collecting voice mail messages is as simple as attaching a recording device to a telephone handset and playing back each of the messages stored in the mailbox. Simple in-line signal pickups are available for under $20 from electronics stores (e.g., Radio Shack part no. 43-1237). Another simple method is to hire a court reporter or other neutral observer to transcribe voice mail messages as they are played back. Transcripts cannot capture nuances such as emotions and tone of voice, but they do provide immediate searchable text. These low-technology, labor-intensive information collection approaches do not scale well to harvesting information from more than a few voice mail accounts, but they do offer simple ways to access this information with minimal interference with ongoing business operations.

Collecting voice mail messages can also be relatively straightforward in businesses that have integrated voice mail with their e-mail systems. Since these systems automatically export voice mail messages as .WAV files and include some phone call metadata, it may be possible to collect extensive voice mail archives without needing to access the voice mail system itself. On some systems, it remains possible for a voice mail system to retain messages even after its e-mail counterpart has been deleted, so it is important to fully understand a specific system before relying on e-mailed voice mail messages as a comprehensive archive. In addition, harvested voice mail messages must still be transcribed before their relevancy can be assessed. Still, compared to the challenge of navigating a proprietary voice mail system, this remains a less burdensome method of accessing this information.

Voice mail messages are least recoverable from a system's emergency backup data tapes. Backup systems further compress the already-proprietary (and likely encrypted) active data stored on a voice mail system. Restoring this data generally requires a second, identically configured server and appropriate backup software to restore the tape-and still requires expert manipulation of the restored server to begin the tedious process of identifying and exporting raw data files. Absent a strong showing of relevance and substantive value, few courts would likely find that a producing party would have to go to this effort to retrieve voice mail messages.