🔊 Voice AI
12 appsNarrating a script, dubbing a presenter and answering a live phone call are different voice workflows. Decide whether interaction must happen in real time and what authorization exists for the voice or likeness. The inventory spans both rendered media and live agents.
How to choose
Compare pronunciation control, latency, consent and handoff behavior. Test a correction, an interruption and an unsupported request using authorized material.
Build or buy: where the scope changes
- 0 focused build
- 3 partial replacement
- 9 consider alternatives
A script-to-audio queue can use an existing provider. Natural live turn-taking, telephony integration and responsible likeness handling need additional engineering and review.
Existing implementation guides
- HeyGen implementation guide — Turn user-authored text into a clearly labeled avatar-style video assembled from user-owned media with a local model, never cloning real people, and retain provenance for every output.
- Thrumble implementation guide — Answer one inbound number with a realtime STT → LLM → TTS agent, look something up, and log the call outcome.
Explore the recorded alternatives
These comparisons cover specific workflows; they do not imply feature parity with every app in this category.
- TTS WebUI — compared in the catalogue with PlayHT, Murf, LOVO.
- Voicebox — compared in the catalogue with PlayHT, Murf, LOVO.
- Balabolka — compared in the catalogue with LOVO, WellSaid Labs, NaturalReader.
- LOVO alternatives
- NaturalReader alternatives
- WellSaid Labs alternatives
Related categories
Category guidance reviewed against the existing catalogue. Counts and verdicts come from the listed app records; product capabilities, prices and setup requirements retain the dates and sources on each guide. These are scoped assessments, not claims of hands-on vendor testing.