The Anatomy of Sound: How Voiceover, Score, and Songs Divide the Work in Short Films

People making short films usually put most of their energy into the visuals. Only when they watch the finished piece do they realize that the issue affecting the viewing experience most is sound. Sound is easy to overlook because it is not as "visible" as the image. Yet it is precisely the channel that acts most directly on emotion—a film with strong visuals but crude sound will immediately feel "amateur" to the audience, even if they cannot say exactly what is wrong. To handle sound well, the first step is to distinguish what parts it is made of.

Three Types of Sound, Three Responsibilities

Sound in short films can roughly be divided into three types, each with a completely different responsibility.

1. Voiceover: Carrying Information

Voiceover (dialogue) is the only one of the three that carries semantic information. Through it, the audience learns what the characters say and how far the story has progressed. Its core requirement is clarity:

• In terms of volume balance, voiceover should usually sit above the other sound elements
• In terms of rhythm, the speaking speed should match the character’s emotional state rather than use one uniform pace
• In terms of voice quality, it should fit the character’s gender, age, and personality setting

In terms of technical implementation, dialogue voiceover in NadouPro can be generated through text-to-speech: enter the lines, voice quality, and performance settings, and output the voiceover audio. If the voice quality of an existing recording needs to be adjusted, voice conversion can output a changed voice based on the original human voice audio and the target voice quality. In the material description, it preserves the content and prosody, which means what changes is the voice quality, not the way of speaking. There is one easily overlooked judgment here: voiceover is not just "reading lines." The same line, read with different speeds, pauses, and stresses, may mean something entirely different. So when setting performance parameters, "how to read it" needs more careful thought than "what to read."

2. Score: Carrying Emotion

Score (BGM) does not carry information; it acts on feeling. Its core traits are continuity and background presence: it lies beneath the image, and while the audience usually does not actively notice it, it is constantly influencing emotional judgment. There are two ways to source music:

• One is to generate accompaniment or background music based on a style, emotion, or scene description;
• The other is to match music automatically according to the video content.

NadouPro’s accompaniment and background music generation follows the first approach: enter a style, emotion, or scene description, and output accompaniment or BGM audio; automatic music matching is suitable for scenarios where music is scored according to video content. The former offers more precise control, while the latter is more efficient. There is a common misconception about score that is worth mentioning separately: score does not necessarily need to be "pleasant to listen to"; it needs to be "appropriate." A melody that sounds beautiful on its own can ruin a scene if it is emotionally misaligned with the image.

3. Songs: Carrying Theme

The difference between songs and score is that songs have lyrics and vocals, so they carry both information and emotion. Because they contain explicit text, their function is closer to a "thematic statement": they usually appear in the opening, ending, or sections with special meaning. NadouPro’s song generation can produce a complete song based on inputs such as lyrics, style, emotion, and instrumentation. When specific lyrics are needed, they can also be provided separately through lyrics input. One point to note when using songs is that they are more "attention-grabbing" than score. Once sung vocals appear, the audience’s attention is immediately drawn to them, and the image takes a secondary position. So songs are usually used in places where the image does not need to carry the main information.

The Logic of How the Three Divide the Work

Putting the three types of sound together makes their working relationship clear: this division determines the basic principle of mixing—voiceover in the foreground, score in the background, and songs depending on the situation. If the volume relationship among the three is reversed—for example, if the score overwhelms the voiceover—the audience will feel that they "can’t hear clearly but don’t know why." This is the most common sound problem.

Where Sound Planning Should Begin

One practical suggestion is: do not leave sound design until the end. The reason is that the dependency between sound and image goes both ways. Some sections have sound that "works with the image" (for example, music following the emotion), while some images "work with the sound" (for example, adjusting shot length to match the rhythm of a line). If all sound is handled only after editing is complete, there is only one-way adaptation: the image has already been fixed, so the sound can only accommodate it. But if sound is planned in advance, the two can be coordinated both ways. The specific method is to mark the sound intention for each shot during the storyboard stage—whether the shot has dialogue, whether it needs music, and whether silence should be left. Then in post-production, sound handling becomes "executing the plan" rather than "rethinking the concept."

Common Misconceptions

• Misconception 1: Treating sound as the final step. Sound and image need coordination; doing it late means you can only accommodate what is already there.
• Misconception 2: Filling the entire film with score. Without silence, there is no tension.
• Misconception 3: Reversing the volume relationship between voiceover and score. The audience cannot hear the lines clearly, but cannot identify the reason.
• Misconception 4: Ignoring the match between voice quality and character. If an adult male character is given a teenage voice quality, the audience will immediately be pulled out of the story.
• Misconception 5: Using songs in sections where the audience needs to pay attention to the image. Vocals take attention away, so placement matters.
• Misconception 6: Letting the three types of sound operate independently. If voiceover, score, and songs are completed separately, style conflicts can easily appear.

Finally

The composition of sound is not hard to understand; what is hard is realizing that it needs to be designed. Images are explicit, and whether they are good or bad is visible at a glance; sound is implicit, making people immersed when it is done well and uncomfortable when it is done poorly, but in both cases it is hard to point out clearly. This is also why sound is often left until the end and handled hastily. Plan voiceover, score, and songs as three independent layers of expression rather than simply "adding some background music," and the overall quality of the work will feel noticeably different. After all, audiences can tolerate flaws in the image, but it is difficult for them to tolerate rough sound.

Contains AI-generated information that may be incorrect. Please verify independently.