Audio drama production is the work of writing, voicing and editing the episodes of an audio series. It is a part of every audio series app. India's audio apps produce thousands of hours of new episodes each month. So production is one of their largest costs, and it needs some of their largest teams.
Three kinds of worker make an audio series. Writers write the scripts. Some series are new. Others are adapted from a popular story on an online story writing platform. The writer turns its chapters into scenes that work without pictures. So every action has to be heard, in a line of dialogue or in a sound. Voice artists read the scripts. A series can use one narrator who voices every character, or a full cast, with a different artist for each character. Sound editors clean each recording, cut mistakes and long pauses, and add background music and sound effects, such as rain or footsteps.
Imagine a production team in Mumbai starting a new Hindi thriller of 300 episodes, each about 12 minutes long. That is 60 hours of audio. The writer delivers two scripts a day, of about 1,600 words each. A voice artist records each episode in about an hour, often in a small studio at home. A sound editor then spends about two hours on each episode. The app releases two episodes a day, so the whole series takes 150 days to release. Listeners expect a new episode every day, so the team works a few weeks ahead of the release schedule.
AI models now help at several steps. A large language model can draft an episode outline, or turn a written chapter into a first script, which a writer then corrects. Speech and voice AI can voice a narrator. It can also voice the same series in Tamil or Telugu from a translated script, in hours instead of weeks. In this example, a Tamil version of the thriller could start only a few days after the Hindi one. An editor still listens to every episode voiced by AI, because a model can mispronounce names or read an emotional scene in a flat voice.
This raises two questions. The first is consent. A model should copy a voice artist's voice only with the artist's written permission, and with payment. The second is whether listeners will accept voices made by AI for stories they care about. Both questions are discussed in AI-generated content.