Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

feat: tts batch stream generate waveform #74

Open
wants to merge 40 commits into
base: main
Choose a base branch
from
Open

feat: tts batch stream generate waveform #74

wants to merge 40 commits into from

Conversation

weedge
Copy link
Contributor

@weedge weedge commented Feb 20, 2025 •
edited
Loading

feat:

  • add tts static batch stream generate waveform
  • add tts stream inference
  • add stream-factor cmd param and TTS_TEXT env param
  • add flow (sub audio vq tokens-> sub mel), hift (sub mel -> sub waveform) session cache
  • add token overlap to gen
  • add max_stream_factor for dynamic batch stream
  • add stream_factor stream_scale_factor max_stream_factor token_overlap_len for tts inference stream params

colab (T4)：https://github.com/weedge/doraemon-nb/blob/main/tts_step_audio_stream.ipynb

weedge added 10 commits February 19, 2025 23:02
@weedge
feat: add flow_cache and hift_cache
b4f06f3 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add tts static batch stream generate waveform
1b096e8 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add tts stream inference
a79c5af 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add tts static batch stream to merge
c867728 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
streamer skip prefix prompt
46f22e9 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: step1lm generated token ids - 65536 to get vq codes
332b5f9 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: lock
c5314f2 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add stream_factor cmd param and TTS_TEXT env param
0634c43 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add stream-factor cmd param and close debug print
de512a4 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: fast path to check params
a62ae84 
Signed-off-by: weedge <weege007@gmail.com>
This was referenced Feb 20, 2025
Open
Open
weedge added 18 commits February 20, 2025 21:54
@weedge
merge stream
2813963 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix typo
e9d1c7c 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix typo
075feef 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: tts instance share streamer -> gen session streamer
7ce0a88 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
merge
e970f3f 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
add speaker_file_path
90a23e0 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
add speaker_file_path
b337668 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
add speaker_file_path
1d8cb6e 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
add speaker_file_path
68752ba 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add modal run step tts/voice
a833d67 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
Merge branch 'feat/flow_hifi_cache' into feat/dev
698f55d
@weedge
add ThreadSafeDict for tts lm,flow,hift session gen
3f17784 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: token_overlap_len
73fb149 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: token_overlap_len
b4a2cf6 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
remove print
ec9d5c6 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add token2wav with flow hift session
71a83d7 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
remove python-dotenv
a25e16d 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix token2wav
8ba35a1 
Signed-off-by: weedge <weege007@gmail.com>
weedge added 11 commits February 23, 2025 20:21
@weedge
fix token2wav
733e052 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix token2wav
42ff2bf 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix token2wav mel
73897a3 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: flow infer return mel float
37c6eee 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
fix: lock
5a426dc 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
token2wav load to cpu
8a2b3f3 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add token overlap to gen
bb0e08e 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
feat: add max_stream_factor for dynamic batch stream
9866e8c 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
change device_map default auto
6b56e20 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
change max_batch_size
680e934 
Signed-off-by: weedge <weege007@gmail.com>
@weedge
add: stream_factor stream_scale_factor max_stream_factor token_overla…
376d762 
…p_len for tts inference stream params

Signed-off-by: weedge <weege007@gmail.com>
@weedge weedge changed the title feat: tts static batch stream generate waveform feat: tts batch stream generate waveform Feb 24, 2025
@weedge
fix tts_inference_stream params
7f623b0 
Signed-off-by: weedge <weege007@gmail.com>
@weedge weedge mentioned this pull request Feb 24, 2025
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@weedge