Common Voice
PulseAugur coverage of Common Voice — every cluster mentioning Common Voice across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Programmer quits due to mental pressure, focuses on language translation for Common Voice
The author has decided to stop programming due to mental pressure and is now focusing on translating for the Common Voice crowdsourcing project. They are translating into the Bandari dialect of Achomi, aiming to include…
-
Pruning Whisper-small improves ASR accuracy by acting as a regularizer
Researchers have demonstrated that neural network pruning can act as a regularization technique for Automatic Speech Recognition (ASR) systems, rather than just a method for compression. By analyzing the sensitivity of …
-
BuzzASR releases 100+ language-specific speech models, outperforming Whisper
Researchers have developed BuzzASR, a suite of over 100 specialized speech recognition models fine-tuned for individual languages. These models are based on the Whisper architecture and significantly outperform the gene…
-
New TTS pipeline enhances ASR systems with phoneme-based augmentation
Researchers have developed a unified pipeline for generating synthetic speech to improve automatic speech recognition (ASR) systems. This pipeline utilizes a multilingual text-to-speech (TTS) model, F5-TTS, with languag…
-
Voice cloning tech repurposed for anonymization, raising security concerns
Researchers have found a way to repurpose the XTTSv2 voice cloning model for speaker anonymization without retraining. This model, trained on extensive speech data, can preserve prosodic structure independently of speak…
-
New benchmark compares multilingual models for Nepali ASR
A new study benchmarks six multilingual pre-trained models for Nepali Automatic Speech Recognition (ASR) using a standardized fine-tuning protocol. The research found that Whisper-Large-v3-Turbo and IndicWav2Vec perform…
-
OpenAI launches new context-aware transcription models for API
OpenAI has launched two new API transcription models: GPT-Live-Transcribe for real-time, low-latency audio, and GPT-Transcribe for asynchronous tasks. Both models are designed to better understand context, leading to im…
-
ai-sage releases GigaAM Multilingual speech models
ai-sage has released GigaAM Multilingual, a family of Conformer-based foundation models. These models, available in 220M and 600M parameter variants, have been pre-trained on over 2 million hours of speech data spanning…
-
New ASR methods tackle compute scaling and multilingual evaluation
Researchers are developing new methods to improve automatic speech recognition (ASR) systems. One approach, LARM, uses a depth-conditioned looped Transformer to allow for adjustable test-time computation, achieving perf…
-
New SBPN model boosts Nigerian language ASR via knowledge distillation
Researchers have developed a new multilingual Automatic Speech Recognition (ASR) framework called Sometin Beta Pass Notin (SBPN) to improve performance for Nigerian languages. The framework uses a two-stage knowledge di…
-
Coqui and Hugging Face advance open-source voice cloning with consent
Coqui, a speech technology startup, is making significant contributions to open-source speech technology and voice cloning. The company focuses on open access models and data, enabling the creation of emotionally resona…