UPD: At the time i did this comparison there were no OpenAI Whisper. Now i use it. And for recognizing voice commands for my smart home i use Orange Pi with Vosk.
Когда я проводил тестирование, OpenAI Whisper еще не существовал, теперь я пользуюсь им, а команды для умного дома распознает Orange Pi с Vosk.
I tested the most popular russian speech recognizers.
Я протестировал наиболее популярные сервисы и библиотеки распознавания русской речи
Video with results: https://www.youtube.com/watch?v=5aXNzigJ0c4
YouTube: https://www.youtube.com/c/MautozTech
YouTube channel about programming: https://www.youtube.com/@ESPdev
Vosk API:
https://alphacephei.com/vosk/
https://github.com/alphacep/vosk-api
https://alphacephei.com/ru/
Google Speech Recognition:
My GoogleContinuousRecognition App: https://github.com/Mike-Kuznetsov/GoogleContinuousRecognition
Android: https://developer.android.com/reference/android/speech/SpeechRecognizer
Google Cloud Speech-to-Text: https://cloud.google.com/speech-to-text
Python: https://pypi.org/project/SpeechRecognition/
Yandex SpeechKit: https://cloud.yandex.ru/services/speechkit
WebSpeech API:
My AndroidWebSpeech App: https://github.com/Mike-Kuznetsov/AndroidWebSpeech
Usage example: https://codepen.io/harryheman/pen/VwppNgp
Docs: https://developer.mozilla.org/ru/docs/Web/API/Web_Speech_API
Mail.ru (VK) Cloud Voice: https://mcs.mail.ru/cloud-voice/
Amazon Transcribe: https://aws.amazon.com/ru/transcribe/
Wav2vec:
Russian: https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-russian
wav2vec tutorial: https://www.youtube.com/watch?v=Ivz-E2eBUZM
Github: https://github.com/pytorch/fairseq/tree/main/examples/wav2vec
Red - Not tested;
Yellow - Tested on phone;
White - Tested on PC;
Columns in english: Service or toolkit name / Price for an hour / Offline mode / Quality of recognition with 3 different volume levels
I tested some services after i made this comparison, here're the results:
Sber Smartspeech - 0 0 0
Tinkoff Voicekit - 0 3.5 4.5
NLab Speech ASR - 0 3 4
Silero - NT NT 3