Towards Quantifying Benchmark Optimization in ASR Models
Paper • 2608.19936 • Published • 8 •
3
Ojin AI
OjinWorld
AI & ML interests
Real-time voice and face models. Interested in turn-taking and endpointing, streaming ASR, expressive TTS, and talking-head generation. Berlin.
Recent Activity
liked a model about 20 hours ago
distil-whisper/distil-large-v3 liked a model about 20 hours ago
openai/whisper-small liked a model about 20 hours ago
pyannote/segmentationOrganizations
Higgs Audio Studio — portable 1-click local build (Windows, + Linux/macOS via Pinokio)
🔥 5
2
#15 opened 2 months ago
by 
nerualdreming
Built a desktop TTS studio GUI for Supertonic 3 - VoxCraft
1
#26 opened 2 months ago
by 
gsk158
🤗[Made with OmniVoice] "A Million Little Fibers" Project 🤗
5
#28 opened about 2 months ago
by 
projectlosangeles
commented a
paper 1 day ago
Built an Android app that runs Supertonic 3 fully on-device - Voxtonic
1
#29 opened 1 day ago
by 
GHyeok
Built a Linux dictation tool on this model, word by word typing on GNOME Wayland
1
#22 opened 2 days ago
by 
scottnelsonathome
Missing quantizations ggml-large-v3-q8_0.bin ggml-large-v1-q8_0.bin ggml-large-v1-q5_0.bin
1
#32 opened 27 days ago
by 
RaivisDejus
Hosted API for VibeVoice-ASR
2
#27 opened 3 days ago
by 
Neil233
Specifying the target language during cloning
1
#20 opened 21 days ago
by 
Markobes
Certain words are repeated many times more than they are repeated by the speaker
1
#18 opened about 2 months ago
by 
VM1991
Streaming microphone transcription emits Unicode replacement characters for Chinese text
1
#49 opened 7 days ago
by 
zhengsihua
Benchmark result: OmniVoice across EN/DE/AR/ES/ZH
1
#30 opened about 2 months ago
by 
aufklarer
Question: edge/mobile deployment — anyone tested?
1
#11 opened about 2 months ago
by 
3morixd
latency profiling script
1
#30 opened 10 days ago
by 
Okietrained
Streaming question
2
#18 opened 11 months ago
by 
koifish12
What's the best way to run this in production for online serving?
2
#45 opened about 2 months ago
by 
fikrikarim
SNAC output quality with high-pitch speech
3
#23 opened about 1 year ago
by 
matthen
Reducing latency by lowering sample rate
2
#24 opened about 1 year ago
by 
liambarryarm
Nemotron 3.5 ASR Streaming 0.6b in native C++/GGML. RTF = 0.0066 on RTX 5090
🔥 3
2
#24 opened about 1 month ago
by 
audio-cpp