-
96
StyleTTS2 ukrainian demo
πStyleTTS2 trained on ukrainian dataset
-
4
RAD-TTS++ Ukrainian (HiFiGAN)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
-
2
RAD-TTS++ Ukrainian (Vocos)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
-
1
RAD-TTS++ Ukrainian (BigVGAN v2)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
AI & ML interests
We are driving innovation in Ukrainian speech technology πΊπ¦
Recent Activity
View all activity
Organization Card
Hello or ΠΡΠΈΠ²ΡΡ [preΛuΜ―Κ²it] π
The Speech-UK initiative is dedicated to developing and providing speech technologies for the Ukrainian language, available globally.
Our activities include:
- Dataset creation and curation;
- Model training and development;
- Tool creation and open-source repository maintenance.
Community:
- Discord: https://bit.ly/discord-uds
- Speech Recognition: https://t.me/speech_recognition_uk
- Speech Synthesis: https://t.me/speech_synthesis_uk
π€ Support our work using Monobank or give us a Star on https://github.com/egorsmkv/speech-recognition-uk
-
2
Speech-to-Text for Ukrainian v2.1 (W2V-BERT 2.0)
πConvert Ukrainian speech to text
-
Yehor/w2v-bert-uk-v2.1
Automatic Speech Recognition β’ 0.6B β’ Updated β’ 668 β’ 2 -
Yehor/whisper-large-v3-turbo-quantized-uk
0.2B β’ Updated β’ 92 β’ 1 -
Yehor/whisper-large-v2-quantized-uk
0.3B β’ Updated β’ 10
-
96
StyleTTS2 ukrainian demo
πStyleTTS2 trained on ukrainian dataset
-
4
RAD-TTS++ Ukrainian (HiFiGAN)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
-
2
RAD-TTS++ Ukrainian (Vocos)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
-
1
RAD-TTS++ Ukrainian (BigVGAN v2)
π§Use RAD-TTS++ model to synthesize text in Ukrainian
-
2
Speech-to-Text for Ukrainian v2.1 (W2V-BERT 2.0)
πConvert Ukrainian speech to text
-
Yehor/w2v-bert-uk-v2.1
Automatic Speech Recognition β’ 0.6B β’ Updated β’ 668 β’ 2 -
Yehor/whisper-large-v3-turbo-quantized-uk
0.2B β’ Updated β’ 92 β’ 1 -
Yehor/whisper-large-v2-quantized-uk
0.3B β’ Updated β’ 10
spaces
5
Building
1
KenLM UI
π
Score texts and build KenLMs
Sleeping
Evaluate ASR outputs
π
Calculate WER/CER values from JSONL files made by ASR models
Sleeping
1
See ASR outputs
π
See generated JSONL files made by ASR models as a dataframe
Running
6
VAD with MarbleNet
π£
Voice Activity Detection using MarbleNet model
Sleeping
1
Text Normalization
β
Verbalization and Text Cleaning for Text-to-Speech models
models
0
None public yet
datasets
19
speech-uk/youtube-videos-enriched
Viewer
β’
Updated
β’
870
speech-uk/youtube-videos
Viewer
β’
Updated
β’
4.78k
β’
97
β’
1
speech-uk/youtube-channels
Viewer
β’
Updated
β’
1.25k
β’
167
β’
3
speech-uk/voiced-news-comments
Viewer
β’
Updated
β’
2
β’
51
β’
1
speech-uk/generated-news-comments
Viewer
β’
Updated
β’
2.2k
β’
57
β’
1
speech-uk/text-to-speech-phonemized-sentences
Viewer
β’
Updated
β’
9.72k
β’
45
β’
1
speech-uk/text-to-speech-sentences
Viewer
β’
Updated
β’
9.72k
β’
45
β’
1
speech-uk/test-ai-house-mp3
Viewer
β’
Updated
β’
37
β’
37
β’
1
speech-uk/test-ai-house-podcast
Viewer
β’
Updated
β’
35.4k
β’
72
β’
1
speech-uk/test-ml-datahoarding
Viewer
β’
Updated
β’
1.22k
β’
55
β’
1