Randomized Crowd

The first thing to start with is a base randomized crowd to serve as a backdrop. Here we have a simple crowd loop with some randomized one-shots over the top (the location specific chants could be swapped out depending on the stadium using a -Switch- and <Set Int Parameter>) as seen in Chapter 06.

Crowd Excitement Levels

Crowd sounds don’t typically start and stop immediately—they build and ebb, so they are most suited to using curve graphs to control the volume of the different elements (the principle of using a value or parameter to read through volume curves to bring crowd elements in and out could be applied to many other crowd situations, for example the panic level variable of a street crowd when shooting occurs).

The systems for the three main crowd elements, {Crowd_Ambience}, {Crowd_Excitement}, and {Crowd_Frenzy}, are governed by the player’s position on the pitch (based on the premise that, as the player moves closer to the goals at either end, the corresponding supporters will get more excited by the promise of a goal being scored). This is illustrated in the following diagram, with general crowd sound represented in blue, crowd excitement in yellow, and crowd frenzy in red.

An <Event Tick> gets the distance between the player and a trigger box that’s been placed at the right hand (red) goalmouth. This value is normalized (i.e., converted into a range between 0.0 and 1.0) and is then used to read through a set of curves that set the volume multiplier of the different elements.

For example the away team excitement rises when the ball is in play near to their goalmouth and becomes frenzied when very close.

When a goal is attempted, we can simply play the {Miss} or {Goal} waves over the top, and the player is then reset to the middle of the pitch, bringing us back to the {Crowd_Ambience}.

Although the two crowd sounds for the {Crowd_Excitment_Cue} are relatively short (14 seconds and 11 seconds), a sense of repetition is avoided through the use of volume envelopes.

In addition to the differing lengths of the loops themselves creating new combinations through their asynchronicity these long looping volume envelopes (of 180 seconds and 145 seconds) help to further vary the resulting sound.

Color Commentary

As a useful starting point, we’ve used the ambient crowd curve (that is up when the player is in the middle of the pitch) to control when the color commentary might come in since there’s less likely to be any crucial action in this pitch position.

The readout of the general crowd curve is checked every tick to see if it is greater than or less than 0.5 (i.e., if the player is in the center of the field). If the player is indeed in this position, then the <Gate> is opened that allows the <Retriggerable Delay> through (the <Random Float Float In Range> gives us a random delay between 10–17 seconds). The <MultiGate> just cycles through its outputs each time it is triggered to select a different color comment type (1—Weather, 2—The team’s current form, and 3—Injuries).

This is used to <Set Integer Parameter> to control the -Switch- for the color commentary.

Concatenated Dialogue for Play-by-play Commentary

There are certain scenarios (and sports games are one) where simply recording all of the potential dialogue lines becomes a practical impossibility. In the English Premier League, there are 20 teams. In a season each team will play each other team twice, once at home and once away, giving them 38 games. As an example at the start of a match we might want a commentator to say:

In order to have this sentence stating each of the possible matches in the football / soccer season, we’d need to record this 760 times. If we now acknowledge that we have 11 players on each team (not including subs!) and each player might intercept a ball that has been passed by another player (or pass to, or tackle, or foul), then you can appreciate that the number of potential dialogue lines becomes huge.

From the sentence above, you can see that it’s only actually certain words within the sentence that need replacing. We could keep the sentence structure the same each time but just swap in or out the appropriate team names and the words “home” or “away.” Sentences usually comprise one or more phrases. By using a phrase-based approach, you can create a compound sentence by stitching together an opening, middle, and end phrase. This method is called concatenation or stitching. We’ve already come across the -Concatenator- node in the Sound Cue that will string together its inputs end-to-end, and this is what we want to happen to our sentence elements.

In our example there are two teams:

Team A

• 0 Lievsay
• 1 Freemantle
• Van der Ryn
• Boyes

Team B

• 0 Thom
• 1 Burtt
• 2 Murch
• 3 Rydstrom

The first concatenation system takes the player who has scored as a variable to control the switch inside the cue {Soccer_Concat_A}.

This gives us the dialogue line:

Note where we have decided to split the dialogue. Rather than “Spectacular goal from…”/“Murch,” etc., our stitched elements include the “from.” If you say the sentence, you can feel that the word “from” and the player name often flow into each other, so there’s no natural break—making stitching hard and most likely giving a clumsy outcome. Between “goal” and “from,” there’s a more natural gap, although very slight, making this a better place to cut and stitch.

Our two interception cues use the same approach ({Interception} and {Given_it_away}).

When the player loses the ball to the opposition, we have version A:

Version B reverses the order to give the player who lost the ball first, and there’s a variation in the following line:

There are two randomly chosen concatenation systems for a goal score. The first operates in the same way as the interception system, adding the player name to the end of a line, and the second goal concatenation system actually inserts the player name in the middle of a phrase:

{Soccer_Concat_B}.

The final play-by-play system comments on the passing of the ball from one player to another, this time taking the two players involved to determine the -Switch- position within the cue {Pass}.

As noted above, the importance of the play-by-play frequency outweighs that of the color commentary, so in this first version of the level (Soccer01A), we’ve implemented a passive Sound Mix that automatically ducks out any color commentary if a play-by-play line occurs.

Considerations for Concatenated Dialogue

Concatenated dialogue can work well, but the challenge is to get the flow of pitch and spacing over the words feeling right across all versions. It is often the spacing of words that stops it from sounding natural and human (if you want to hear some examples of bad speech concatenation, try phoning up your local utility service or bank). You can also appreciate how a recording session for a concatenated system would differ from a normal session in that you’d want to construct a script so that you get the maximum amount of words you want in the least amount of time, but not have the chunks in such isolation as to lose the natural delivery of the words as they would be in the final context (and as you can imagine, this approach also presents a nightmare for localization where you are dealing with different languages).

Although functional, you can hear that in many cases it does not always sound very natural. In a real match these comments would also vary in inflection depending on the circumstance, for example a pass of the ball a little way down the field would be different from a pass that’s occurring just in front of the goal of the opposing team with only thirty seconds left of the match. It might be that you record two or three different variations to be chosen depending on the current game intensity level. The problem with more excited speech is that we tend to speak more quickly. Even in normal speech, our words are often not actually separate but flow into each other, and this is exacerbated by speed. You recall that we discussed the importance of judging where to cut your sentences, looking out for opportunities where you can stitch seamlessly (for example on stopped consonants such as t, d, p, b, k, and g where the mouth is closed—therefore creating a silence). More seamless stitching can be achieved, but it requires a deep understanding of aspects of speech such as phonemes and prosody that are beyond the remit of this book.

Your system also needs to show an awareness of time and memory by suppressing certain elements if they repeat within certain timeframes. In natural speech we would not continually refer to the player by name but would, after the first expression, replace the name with the pronoun “he” or “she.” However if we came back to that player on another occasion later on, we would reintroduce the use of their name.

All these things would be significant challenges in themselves, but are heightened by our innate sensitivity to speech and language and the huge amount of time that many players spend playing these games. They may be played regularly over a series of months or years during which players are listening to the commentator dialogue all the time. Rarely are game audio systems exposed to such scrutiny as this.