Dub Together karaoke refers here to singing while redubbing a video clip, a theme found in a verified community video's title. The official game still describes two modes, Parallel and Character, rather than a separate karaoke or rhythm-game mode. Its September 8 update directly addresses singing synchronization and introduces a 150 ms VOICE TIMING default. A licensed song catalogue, dedicated karaoke pack and exact scenes from that community video are 待确认.
Understand the singing use case
Dub Together replaces a clip's voices with recordings made through the players' microphones. Singing can be approached as a kind of vocal performance over the footage, with the same need to fit delivery to the picture and audio. The developer's September 8 note says synchronization problems were more obvious when singing than in normal speech. That makes voice timing the clearest documented starting point for this particular use case.
The community karaoke video provides a showcase destination for the theme, but the verified detail is its title topic. It does not establish an included song, official music licence or a specific downloadable pack. Use it as a community example when deciding what sort of performance interests you. The exact scene, lyrics and creator workflow behind that video remain 待确认, so this guide does not supply them.
Choose separate takes or shared roles
Parallel gives every player the same clip and lets them record independently without hearing the other versions. The group then watches the complete performances in turn and votes for a favourite. That structure can suit friends who want to hear different vocal interpretations of the same material. It is still the documented dubbing mode, with a preference vote rather than a verified singing-accuracy score.
Character assigns a performer to each role and brings the dialogue into one shared version. Players follow the lines together, and an unassigned role keeps its original voice. This mode needs at least two players and has no vote at the end. Before choosing it for singing material, inspect the actual roles and lines in the pack instead of assuming that a duet, harmony system or special song arrangement is included.
Read the September timing change
The September 8 announcement explains that recorded voices could previously feel slightly out of sync with the video's music. It says that delay is now compensated and gives the default VOICE TIMING value as 150 ms. The same announcement says the endings of recorded lines are now captured properly. These are the published changes relevant to singing, rather than independent measurements from a microphone or headset test.
The direction of the setting matters: a higher number makes the recorded voice start earlier, and a lower number adds delay. Do not read a larger value as a longer wait before the voice begins. The developer says players can fine-tune the setting for their audio setup. It does not publish one universal setting that guarantees perfect synchronization for every arrangement of speakers, headphones and microphones.
Adjust timing with a clear direction
Start by understanding what you hear in the recorded result, then relate that observation to the documented adjustment direction. If you decide the voice should begin earlier, the announcement describes increasing VOICE TIMING as the relevant change. If you decide it should be delayed, it describes decreasing the value. This is an explanation of the setting, without claiming that any particular device has been tested or needs a prescribed correction.
The September 15 update makes VOICE TIMING easier to understand and shows the actual timing adjustment. Refer to the displayed setting when discussing an attempt with your group, because a description such as earlier or later is more useful with a known value. Recordings may be repeated while the round timer runs, which allows another take during that available window. The duration of the timer and a best adjustment step in milliseconds are 待确认.
Keep line endings and effects in mind
Singing stretches sounds differently from ordinary dialogue, so pay attention to the ending of the line when listening back. The September 8 fix concerns proper capture of line endings, without defining a special singing score or note-recognition mechanic. Follow the actual line boundaries provided by the pack and listen to the completed recording. Automatic lyric accuracy, pitch grading and a sustained-note bonus are 待确认 in the official description.
The game documents six voice effects, including Chipmunk, Monster, Echo, Robot and Phone, selectable per line. Effects are applied to the recording, so the group hears the altered version during playback. The unnamed sixth effect is 待确认, and the noise filter is described as removing background noise between sentences. Try an effect for the interpretation you want, while keeping timing and the sound of the chosen effect as separate listening questions.
Find a suitable clip without assuming a catalogue
Steam Workshop contains packs made by players, and the September 6 update adds filters for language, line count and role count. Those filters help you inspect material that fits your group, but they do not constitute a verified official song menu. Read the actual item's description before assuming that it contains the performance you have in mind. The only bundled pack identified in the Steam listing is the Sprite Fright excerpt under CC BY 4.0.
Subscribed packs appear without restarting, and missing round packs download automatically when play begins. That is the documented route for receiving a pack, without a need to invent a separate karaoke download site. A community item being visible on Workshop also does not establish that every included song is officially licensed by Banani. Availability of a named karaoke pack, its exact tracks and its permission status remain 待确认 unless the particular item provides the necessary information.
Prepare singing material in Pack Creator
Pack Creator imports a video, lets you trim the excerpt and detects speech with matching text for the lines. It separates speech, music and background sound, then lets you assign roles and export a playable pack. These are the documented editing steps, but the sources do not guarantee transcription quality for every sung passage. Review the detected lines for your actual footage instead of assuming a finished lyric sheet or a music-specific editing mode.
The September 15 update adds + NEW LINE at the playhead and allows missing source videos to be relinked with FIND VIDEO without losing existing lines. Uploads through Workshop include rights confirmation, and the developer requires permission for other people's film, television or music material. Prepare a clip you have the right to share before considering a public upload. The clips guide explains these creator actions and the difference between a playable pack and a saved performance.
Watch, vote and save the performance
Parallel reveals every version in turn and ends with a favourite vote, while Character plays one shared result without voting. The host's September 15 controls allow rewatching or advancing between versions. Listen to timing, delivery and line endings as practical parts of your own performance review. The group can choose a favourite without a documented numerical score for pitch, rhythm or lyrical accuracy.
Finished dubs can be saved as videos, and the September 15 announcement gives the automatic location as Videos\Dub Together. A verified export format, resolution and dedicated karaoke sharing service are 待确认. Keep the distinction between the footage used for the round and the video of the performance you have just made. For organising the friends who will sing or watch together, use the online session guide.
What players ask
Is karaoke a separate official game mode?
The official description lists Parallel and Character as the two modes. A verified community video title uses a karaoke theme, which establishes a showcase topic. The September 8 update specifically discusses voice timing when singing. A separate rhythm-game mode or official licensed song catalogue remains 待确认.
Which way should VOICE TIMING move?
The September 8 announcement sets the default to 150 ms. A higher value starts the recorded voice earlier. A lower value adds more delay. The notes allow fine-tuning for an audio setup but do not establish one best value for every singer.
Can I download an official karaoke song pack?
No official karaoke pack was established in the reviewed sources, so its availability is 待确认. The named bundled content is the Sprite Fright excerpt by Blender Studio. Player-created packs are shared through Steam Workshop. Check a particular item's content and permissions instead of treating a community video title as a download link.