The Audio tab covers voice, music, sound effects and dialogue. The models offered are mainly ElevenLabs models, routed like everything else through fal.
🔺 There is no "mode" dropdown in this tab: the model you choose determines what you can do and which fields appear. Pick the model first, and the form adapts.
Text to Speech
You enter a text and select a voice from the list the model offers. The model generates an audio file of that voice reading the text, with fine control over stability, speed and, on some models, speaker boost.
- 🎙️ Scene narration, 💬 NPC dialogue, 📚 accessibility (audio versions of written text), 🌍 multilingual dubbing.
Music
You describe in text the style, mood and emotion you want ("epic action music for a fight scene, orchestral, building tension") and the model generates an original track.
- 🎮 Game soundtracks, 📖 narrative music, 🏋️ training ambience, 🎯 audio prototyping.
Sound Effects
You describe the effect you want ("distant thunder rolling closer", "mechanical button click"). Depending on the model, you control the duration (with a slider, and a switch to leave it to the model) and the prompt influence (0 = very free to 1 = very faithful). A loop option produces seamless looping effects.
- 💥 Interaction effects, 🌊 looping ambiences, ⚔️ action effects, 🔔 notification sounds.
Multi-voice Dialogue
With a "conversation" model, a rich text editor replaces the prompt field: you compose the conversation and assign a different voice to each passage using the Voice tool in the editing bar. The model produces one continuous audio file chaining all the lines with the right voices.
- 🗣️ Ready-to-use dialogue scenes, 🎭 audio cutscenes, 📻 podcasts / microlearning, 🌍 multilingual dialogue (via the language code field, where the model offers one).