ChatCut’s inspected music route creates a new instrumental with mureka-9 or a new vocal song with mureka-9.5, one audio asset per job. Describe the music, approve lyrics for a song, then place and trim the completed asset. Exact generated duration is unverified; the route has no duration field. It does not expose Mureka’s separate remix, stem or extension APIs.
Last verified: 2026-10-28. This reference combines official API documentation, ChatCut’s source contract at b3cbb24 and bounded endpoint validation. It is not a new production availability or audio-quality test. See the Chinese translation.
Verified facts and evidence scope
| Control | Native model and ChatCut route | Evidence and source |
|---|---|---|
| Model identity | Inspected ChatCut defaults: mureka-9 for instrumental; mureka-9.5 for song. | ChatCut source review, 2026-10-28; ChatCut music workflow, Mureka instrumental API and song API. |
| Text | Prompt up to 1,024 characters; lyrics up to 5,000. | Source contract plus over-limit validation, 2026-10-28. Validation confirms rejection beyond the caps, not a successful generation at the boundary. Song API. |
| Output count | ChatCut requests one output per job; the native API can return multiple choices. | Source review, 2026-10-28; instrumental API, ChatCut workflow. |
| Duration | No requested-duration field in the inspected tool or generation schemas. Returned duration is measured in milliseconds. A new-song output ceiling is unverified. | Official schemas and inconclusive endpoint probe, 2026-10-28; song API. |
| Existing audio | Mureka documents separate remix, region editing, stems and extension APIs. ChatCut has not exposed those controls in this music route. | Source review, 2026-10-28; Mureka changelog, ChatCut workflow. |
| Credits | Request a current estimate; compare it with the completed ledger charge. No universal cash price per track is verified here. | Pricing evidence reviewed 2026-10-28; ChatCut credit policy, generation credits. |
Choose an instrumental or a song
Use an instrumental for a background bed under dialogue, a podcast opening or a travel sequence. In the inspected tool, a text prompt supplies the direction. Mood, genre, instruments and tempo are creative guidance; they do not establish an exact beat map or a guaranteed result.
Choose a song when the lyrics and singing are part of the piece. Prepare and approve original lyrics before submitting. The route accepts lyrics, a direction prompt, an optional male/female vocal preference and an asset name. A vocal preference is not a guarantee of a particular singer or vocal identity.
The provider’s own song API also documents reference, melody and vocal IDs. Those controls are not exposed by this ChatCut tool. A picture supplied to the Agent may inform a text description, but it is not a native image input to the instrumental or song route. Read generation references before assuming a reference slot applies to every model.
Generate first, then fit the edit
Set the editing target
Choose the clip or range to score and decide whether it needs an instrumental or a song.
Write the direction
For an instrumental, describe mood and instruments. For a song, include approved lyrics and singing direction. Review the credit estimate before submitting.
Review the completed asset
Wait for completion in My Assets and listen to the whole result. Check its actual duration and whether it fits your intended use.
Place and mix
Add the asset to an audio track, trim the passage you need, position it against the picture and adjust volume around speech.
Example direction: “Generate a new instrumental with warm electric piano, soft percussion and a calm mood. Add it under the interview, then trim it to the selected range and keep the dialogue audible.” The selected range is the desired timeline use, not an exact-duration generation setting. The expected editable output is a regular audio clip; assess the generated music before using it.
What the endpoint checks do and do not establish
A bounded test rejected mureka-9.5 on the separate remix endpoint. That route-specific denial does not establish that V9.5 is unavailable for new songs. A stem-separation attempt returned a server error without a completed archive, so successful separation remains unverified. No successful edit, remix, stem or extension result is established by this reference.
The duration probe failed for missing lyrics. It cannot establish whether unknown duration fields would be accepted, ignored or rejected. Do not treat the provider’s remix-input limits as a new-song output limit or promise an eight-minute song. The tool-level absence of a duration control is the relevant ChatCut limitation today.
Check the charge
Use the editor’s current preflight estimate, then verify the settled charge in Credits History → Detail. Public card estimates are rounded and depend on the catalog/cohort; they are not a current account quote, and a supplier cost is not a retail cash price. Credits Policy owns the tariff; Generation Credits explains the estimate and ledger workflow. Agent usage may be a separate charge.
For publishing, review applicable terms and rights to supplied lyrics or references. This page does not establish unrestricted licensing, exclusivity or immunity from platform copyright claims.
Frequently asked questions
Should I generate a song or instrumental?
Which duration is measured versus requested?
How is the charge verified?
Sources and related guides
Official instrumental generation, song generation and changelog: documentation snapshot accessed 2026-10-27, reviewed with the ChatCut contract and validation evidence on 2026-10-28. Provider documentation describes native APIs; it does not prove current ChatCut account entitlement or production success.
Continue with AI music for video, the music and sound workflow, model capability reference or workflow failure recovery.