Long-form Script Workbench
An open-source speech production workbench based on IndexTTS 2.0, integrating long-form script import, character management, reference audio processing, multi-track editing, spatial audio, and project file export.
Timeline (last 90 days)
Bilibili UP host "登录仓" created a long-form script workbench for the IndexTTS 2.0 super-optimized integrated package, and the project is open source.
According to the demonstration, the workbench consumed as many as 1 billion tokens during generation.
The workbench integrates script parsing, character management, audio editing, multitrack mixing, and project export into one complete pipeline.
The workbench supports directly importing long-form script files, automatically parsing the text and categorizing dialogue by character; users can merge character lines and perform other processing, then preview and re-edit by character group.
The workbench can process reference audio, extract clean vocals, remove leading and trailing silence, and trim clips, and it supports extracting vocal audio from video to create voices as well as batch-assigning reference audio by character.
The workbench provides a multitrack editing interface that references CapCut's editing features and uses IndexTTS 2.0's emotion control capability to adjust emotion and regenerate.
The workbench supports exporting project files that can be directly imported into the professional digital audio workstation Reaper for further editing, preserving the separate tracks, time positions, and basic properties of each character's audio.
The workbench installation process is: download the zip package from the GitHub Release page, place it into the IndexTTS 2.0 integrated package directory and extract it to overwrite, run the install script, then double-click the Start launch script to run the project.
The workbench pages include entries such as "Long-form Script" (path Longform), "Timeline Editing" (path Timeline), and Usage Guide (path Guide).
The workbench's stability under very long scripts, consistency of voices across multiple characters, and fineness of emotion control still lack systematic evaluation.
2 more timeline events
All Facts (12)
Bilibili UP host "登录仓" created a long-form script workbench for the IndexTTS 2.0 super-optimized integrated package, and the project is open source.
50%UnverifiedAccording to the demonstration, the workbench consumed as many as 1 billion tokens during generation.
50%UnverifiedThe workbench integrates script parsing, character management, audio editing, multitrack mixing, and project export into one complete pipeline.
50%UnverifiedThe workbench supports directly importing long-form script files, automatically parsing the text and categorizing dialogue by character; users can merge character lines and perform other processing, then preview and re-edit by character group.
50%UnverifiedThe workbench can process reference audio, extract clean vocals, remove leading and trailing silence, and trim clips, and it supports extracting vocal audio from video to create voices as well as batch-assigning reference audio by character.
50%UnverifiedThe workbench provides a multitrack editing interface that references CapCut's editing features and uses IndexTTS 2.0's emotion control capability to adjust emotion and regenerate.
50%UnverifiedThe workbench supports exporting project files that can be directly imported into the professional digital audio workstation Reaper for further editing, preserving the separate tracks, time positions, and basic properties of each character's audio.
50%UnverifiedThe workbench installation process is: download the zip package from the GitHub Release page, place it into the IndexTTS 2.0 integrated package directory and extract it to overwrite, run the install script, then double-click the Start launch script to run the project.
50%UnverifiedThe workbench pages include entries such as "Long-form Script" (path Longform), "Timeline Editing" (path Timeline), and Usage Guide (path Guide).
50%UnverifiedThe workbench's stability under very long scripts, consistency of voices across multiple characters, and fineness of emotion control still lack systematic evaluation.
50%UnverifiedThe consumption of 1 billion tokens indicates that the workbench's compute cost is not low, and local deployment has certain hardware requirements.
50%UnverifiedThe workbench's opening spatial audio feature may be implemented through multichannel rendering or stereo widening, but the original text does not elaborate on the specific implementation or效果.
50%