MOSS-TTS Local v1.5 Realtime Streaming

Realtime streaming decode with Direct Generation, Clone, Continuation, Continuation + Clone, and language tags

Text

Reference Audio (Optional)

Drop audio here
or
click to upload

No reference selected.

Mode with Reference Audio

If no reference audio is uploaded, Direct Generation will be used automatically.

Clone Continuation Continuation + Clone

Language Tag

Optional for v1.5. Set this when the input language is known, especially outside Chinese and English.

Enable Duration Control (Expected Audio Tokens)

Duration control is disabled.

Sampling Parameters (Audio)

temperature

0.13

top_p

0.11

top_k

1200

repetition_penalty

0.82

max_new_tokens

17500

Codec Chunk Frames (0=auto)

032

Initial Playback Delay (s)

seed (-1=random)

-1999999

Enable Streaming Generation

Output Audio

Download final wav

Status

Runtime: checking...

idle

Examples (click a row to fill inputs)

Reference Speech	Example Text