17.3 Running a Recording Session


A successful recording session depends on a certain familiarity with procedure, as well as knowing what to communicate to a voice actor and how.

17.3.1 Procedural Considerations

There are a number of procedural considerations that will make the session run more smoothly and keep the voice actor in good form. Remember that the actor is not a machine but a person, and people sometimes have special personal and environmental needs to perform at their best.

Prepare the Voice Actor

Don't expect your voice actor to show up, receive a script, and immediately start recording prompts without receiving an overview of the application's purpose, goals, or areas of functionality. In addition to furnishing local context for individual prompts, as recommended earlier, it is crucial that you inform the voice actor of the same global issues that were deemed relevant for creating an appropriate, application-specific persona, as outlined in the persona design checklist given in Chapter 6.

Make sure that the voice actor is aware not only of relevant business concerns but also of any information about typical users, such as demographics, whether they will be calling frequently, their likely mind-set, and so on. Get a copy of the recording script and a description of the interface's persona to the voice actor beforehand if possible, a day before the session.

Recording Procedure

Decide on your recording procedure before the session begins, and inform both the voice actor and the sound engineer of your plan. You should decide in advance how many times the voice actor should say each item in the script.

In planning your recording procedure, think about the nature of the content you are recording and how it functions in the application. Suppose you are recording "one minute" up to "fifty-nine minutes," all with the same intonation contour, or a list of more than 100 relatively simple city-and-state names (e.g., "Providence, Rhode Island"). In that case, instruct the actor to simply do one take of each item, adding something like, "Just move on to the next city, unless I cue you for a retake." This procedure will ensure that the recordings are elicited as quickly and efficiently as possible.

This approach is most appropriate when the items to be recorded will not be heard frequently and when the list is lengthy and repetitive, but simple. In this case, it is not worth eliciting more than one token of each city and state unless it is necessary. On the other hand, if you are recording messages that are more complex or more informative, or that will be heard rather frequently, have the voice actor do three to five takes of each one.

It is not advisable to start the recording session with, for example, the welcome prompt, or system greeting. This is the one message that all users will hear and will be the basis for all users' very first impression. You want to make sure the voice actor is adequately warmed up before reading such critical prompts. The same holds true for other high-traffic prompts, such as "What's your account number?" and "Sorry, I didn't catch that," as well as any difficult or tricky messages, including concatenation pieces.

Planning Tips

Keep these guidelines in mind while planning your session:

  • Duration: The session should not exceed four hours in a given day. Vocal production is muscle activity, and muscles get fatigued with repetitive use.

  • Time of day: Voices change throughout the day. If recording will take place over the course of a few days, schedule consistently.

  • Beverages: A note about beverages may seem trivial, but we raise this issue because they directly affect the vocal tract and therefore the sound of your recordings. Make a beverage available, but avoid ice-cold beverages, which constrict the vocal cords. Also avoid dairy and thick juices, which have a tendency to coat the throat and make the voice sound thick.

  • Breaks: Take a short break every hour, unless you and the voice actor feel you're on a roll and don't want to break the momentum.

  • Sitting versus standing: Although this is a matter of the voice actor's personal preference, the recording area should be equipped with a comfortable seat with decent back support. Many voice actors prefer to stand because it facilitates a more energetic delivery and fuller, more controlled involvement of the diaphragm.

17.3.2 Voice Coaching

A smooth-flowing, comfortable dialog between the user and the system depends in part on what the voice actor both knows and feels in the recording studio, and both knowledge and feelings are mediated by the voice coach. The role of the voice coach is therefore critical in the production of the VUI and should not be underestimated.

First, it is the coach who communicates the application's objectives and functionality, user demographics and psychographics, and the context of the prompts to be recorded. In addition, the coach must also be knowledgeable about language in general. For example, the coach must know whether a Wh- question such as "What's your account number?" is intended as a first-time request or as a request for repetition, and must know (at least implicitly) that such a difference determines the question's intonation contour. Much of what the actor feels in the recording studio will depend on the voice coach's social intelligence, interpersonal skills, and emotional sensitivity.

In general, the coach should avoid editing prompts in the studio. If there is a compelling reason to make a change perhaps some wording is clearly wrong when you hear the voice actor create it make sure to document the change and notify both the grammar writer and the dialog designer.

It is unlikely that the prompts will be recorded in the order that users will hear them, so it is important to maintain consistency throughout the recording session. This means consistency in volume, distance from the microphone, energy, and interpretation of the persona. You can prevent most inconsistencies in vocal quality by being mindful of the duration of the session, the time of day it is scheduled, the consumption of certain beverages, and so on, as indicated earlier.

As early in the session as possible, identify one or two key recordings that you feel are especially successful in capturing the desired mood, energy level, and persona. Keep these recordings handy so that you can revisit them throughout the session. If there are any additional recording sessions, start by having the voice actor listen to these key recordings so that he or she can "find the voice" from the earlier session.

The voice coach should try to create a relaxed, pleasant, warm, and supportive environment. Give the voice actor plenty of positive feedback. Communicate praise and encouragement through your own positive body language. Gestures as subtle as a nod or thumbs up communicate reassurance effectively without disrupting the flow of recording.

Be aware of the type and amount of direction you are furnishing the voice actor. Do not overcoach. In other words, try not to give overly explicit, overly technical instructions about how you want things to sound. Otherwise, you run the risk of making the voice actor overly self-conscious, and that can result in awkward or unnatural-sounding prompts. If you are not making progress on a particular item on the script, a good rule of thumb is to move on and return to the problematic item later in the session.

Voice coaches should be aware of the difference between sloppy diction and colloquial pronunciation. Sloppy diction refers to mumbled, slurred, or garbled speech caused either by a habitual way of speaking or an isolated, unintentional speech production slip-up. Sloppy diction should be avoided in speech interfaces because it detracts from clarity and can thus lead to listening comprehension problems. It also gives the impression of low production values and can therefore be damaging to your branding efforts. In contrast, colloquial pronunciations, such as "gonna," "wanna," and "lemme," are everyday, conversational alternatives to the more formal, more careful "going to," "want to," and "let me." Remember that the voice actor's pronunciation should evoke a linguistically and socially familiar persona. Use the pronunciation variant that is appropriate in the context of the dialog and persona.

More than anything else, a good voice coach is a good listener. The voice coach must be listening attentively while the voice actor is recording. The coach should be paying attention to prosody, the quality of vowels and consonants, undesirable mouth noise, and consistency of volume, energy, mood, and persona. In sum, a good voice coach listens on many levels at once and directs the voice actor to produce recordings that will work in context and satisfy the business and user requirements of the application.



Voice User Interface Design 2004
Voice User Interface Design 2004
ISBN: 321185765
EAN: N/A
Year: 2005
Pages: 117

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net