Request for Official Video Generation and Text-to-Speech APIs

Developers
1

Background

We are building an AI short-video advertising generation app on the Anna AI Platform. The app generates complete promotional videos from user input, product information, reference assets, brand guidelines, and campaign requirements.

Its core workflow often needs to generate and assemble multiple types of media content.At the moment, video generation and text-to-speech capabilities can only be integrated through third-party services inside the app’s own Executa / Tool layer.

At the moment, video generation and text-to-speech capabilities can only be integrated through third-party services inside the app’s own Executa / Tool layer.

In the current project, we may need to integrate external providers for text-to-video, image-to-video, digital human generation, voice synthesis, or audio processing. The app itself must handle provider authentication, task submission, asynchronous polling, file downloading, format conversion, retries, fallback behavior, cost control, result normalization, and workspace artifact management.

This approach is acceptable for prototyping, but it is not an ideal long-term solution for a production app running on the Anna AI Platform.

Expected Outcome

We hope the Anna AI Platform can include video generation and text-to-speech as official Host API / Tool API capabilities, with stable API documentation, permission models, asynchronous task management, error codes, usage statistics, workspace integration, and platform governance.

In the short term, a minimal viable set could include:

  • video_generate;
  • video_get_task;
  • video_cancel_task;
  • video_fetch;
  • tts_generate;
  • tts_list_voices;
  • media_probe.

The minimal implementation could initially support:

  • Text-to-video;
  • Image-to-video;
  • Basic duration and aspect-ratio control;
  • Standard text-to-speech;
  • Voice and language selection;
  • Audio artifacts stored in the workspace;
  • Video artifacts stored in the workspace;
  • Unified asynchronous task states;
  • Unified provider errors;
  • Basic moderation;
  • Usage and quota reporting.

In the longer term, the platform could further support:

  • Multi-provider configuration;
  • Automatic provider fallback;
  • Provider routing based on quality, latency, cost, or region;
  • Video quality evaluation;
  • Multiple candidate generation;
  • Character and product consistency;
  • Video extension;
  • First-frame and last-frame control;
  • Digital human generation;
  • Lip synchronization;
  • Multi-speaker speech generation;
  • Voice cloning with consent management;
  • Word-level and phoneme-level timestamps;
  • Speech-to-subtitle alignment;
  • Media transcoding and normalization;
  • Scene-level video generation workflows;
  • Integration with timeline editing systems;
  • Integration with the app workspace artifact system;
  • Generation provenance and synthetic-media disclosure;
  • Platform-level cost optimization and quality benchmarking.

This would significantly reduce the cost for Anna app developers to integrate external video and speech-generation services.

It would also improve the stability, quality, safety, portability, and governability of multimodal AI applications built on the Anna AI Platform.