# AI News, 16 October: Instant Soundtracks and GPT-Live-1.1

ElevenLabs adds video-aligned music and sound effects with separate tracks, while OpenAI rolls out new ChatGPT Voice models and localized voice choices.

Published: 2026-10-16. Updated: 2026-10-16. Author: XANTRANS editorial.

For finished video, separate generated music and effects leave more room to edit the mix. For spoken experiences, the new voice choices are worth listening to in the actual language and account before choosing them for a project.

## Instant Soundtracks puts music and sound effects on the video timeline

ElevenLabs introduced Instant Soundtracks on 15 October. Upload a video to ElevenCreative and the tool generates an original musical score while analyzing the footage for sound effect cues. Its announcement says the effects are positioned at matching moments on the Music Video Editor timeline, where they remain separate from the music. That separation matters when a cut needs a quieter impact, a different transition, or space for dialogue. A single mixed audio file would make those changes harder after generation.

The company describes tighter timing around cuts, transitions and visible actions in short videos. It also says the music can change mood across scenes, rather than hold one texture for the entire piece. For a 30-second product film, an editor could upload a rough cut, inspect where the generated cues land, then lower an effect under the product name and adjust the score before export. That is a practical use of the editing controls, not a claim that the first generated mix will match a director's intent. Listening and revision still decide whether the result works.

The release spans ElevenCreative's Music Video Editor, Creative Engine for ad work, ElevenMusic and the Studio API. The documented video-to-music endpoint accepts one or more videos as a multipart upload. `add_sfx` requests sound cues, `separate_tracks` returns music and effects in a ZIP rather than one mix, and `use_audio_as_reference` analyzes an existing track's genre, tempo, instrumentation, mood and arc to guide new music. All three are optional; the documentation says their defaults are false. That makes it possible to ask for a straightforward soundtrack first, then request separable effects when the job needs a mix pass.

The API documents a maximum of ten videos, 200 MB combined and ten minutes total. It warns that combining multiple videos can lengthen the request, and recommends assembling the cut first where possible. That is a meaningful boundary for anyone considering an automated video pipeline. The announcement concerns uploaded footage, so it does not establish a continuously generated soundtrack for a live video stream. We have not verified a price per finished minute, a particular account's access, a commercial license for any specific input, or the delay on a real project. Teams should check those details before adopting it as a production step.

## GPT-Live-1.1 brings more voice choices to ChatGPT

OpenAI's 15 October ChatGPT release notes say GPT-Live-1.1 and GPT-Live-1.1 Quip are rolling out in ChatGPT Voice. Quip is described as a smaller version of the model. The company says the update brings more natural accents across languages and expands the short phrases Voice uses while it searches. This is a change to the conversation experience: the voice can acknowledge an in-progress search without relying on the same brief cue every time.

The new voice choices include French, German, Hindi, Japanese, Korean and Mexican Spanish options, plus an English voice called Bubbie. A team preparing a multilingual visitor guide could listen to short passages in the languages it actually needs, paying attention to names, pauses and interruptions. A client could compare whether the spoken delivery fits its brand before recording a full script. Such a listening pass is still needed because a list of supported voice choices does not measure clarity, tone or consistency in a particular application.

To choose a voice in ChatGPT, OpenAI directs users to Settings, then Voice. The release says available choices may differ by region and account, and describes the rollout as ongoing. It does not say that GPT-Live-1.1 or Quip is available through the API, nor does it give a latency or price comparison for an embedded voice product. A product team should treat ChatGPT Voice as the announced access path and check separate API documentation before planning an integration. This roundup has not performed an independent listening test, and the release notes do not establish Latvian quality.

## Sources

- [ElevenLabs Instant Soundtracks announcement, 15 October](https://elevenlabs.io/blog/introducing-instant-soundtracks) (checked 2026-10-16)
- [ElevenLabs video-to-music API reference](https://elevenlabs.io/docs/api-reference/music/video-to-music) (checked 2026-10-16)
- [OpenAI ChatGPT release notes, 15 October](https://help.openai.com/en/articles/6825453-chatgpt-release-notes) (checked 2026-10-16)

Prepared with AI assistance from the linked primary sources. XANTRANS has not independently tested these releases.


Original page: https://xantrans.com/blog/ai-news-2026-10-16-instant-soundtracks-gpt-live/