The sound in my head was much better
Imitating a noise feels easy right up until the recording plays.
Before that, the version in your head is convincing. The robot alarm has the right pitch. The cat sounds like a cat. You can even hear the rhythm of the old clip you are trying to copy.
The microphone may have a different opinion. The robot becomes a bicycle horn. The cat sounds mildly offended. The clip is still recognizable, but only if you are being generous.
I like that gap. A bad imitation is not always a failed attempt. Sometimes it is close enough to be satisfying. Sometimes it is much funnier than the sound you meant to make. Either way, playback removes the argument.
That became , a small browser game about listening to a sound, trying to copy it, and hearing the recording before seeing the score.
A sound plays once. You press record and make your version. When the take ends, it plays back. The score comes after that, which seems fair. A number makes more sense when you have just heard what it is judging.
I expected the scoring to be the interesting engineering problem. I did not want to compare recordings by volume alone. Loud does not mean close. One attempt can have the right rhythm but the wrong pitch. Another can hit the note and still start too late.
The browser looks at pitch movement, timing, duration, rhythm, and changes in energy. The useful signals depend on the sound. A short meme clip cares about timing. A machine noise may be easier to recognize from its attack and texture. An animal call often depends on the rise and fall of the voice.
The score is still only an estimate. Microphones pick up the room as well as the player. They notice how far someone is sitting from the screen, and sometimes they record a cupboard opening in the next room. Two people can make the same sound and give the game very different material to compare.
I wanted the result to give the player a hint about what happened. "Your timing was weak" is something you can try to fix. A score of 48, by itself, leaves you staring at the number and wondering what went wrong.
The microphone raised another question: where should the recording go?
Uploading a voice recording seemed unnecessary for a game like this. Mimic Party processes the take in the browser and keeps it on the device. There is no account to create and no remote audio service waiting for a robot impression. The replay can start as soon as the recording ends.
That does not solve every audio problem. A noisy room remains noisy. A microphone can clip or barely hear the player. When a take is too quiet or too short to judge, the game reports an audio problem instead of handing out a bad score for a recording it never captured properly.
Quick mode is for a few minutes. Survival mode gives you three Mic lives and a longer run of references. A weak attempt costs a life, but it does not end the session straight away. This is probably enough structure for a game about making noises at a screen.
The best version is still the one with several people using the same sound. One person aims for accuracy. Someone else adds a pause that was never there. The last person hears both attempts and blames the prompt. Nobody agrees, and someone asks for another round.
The project is small on purpose. It gives you one odd thing to do, then lets you leave. You do not need a profile or a record of every bad impression. Make the noise, listen back, and close the tab if you are done.
评论
发表评论