Yesterday, Daniel Povey, an international expert in speech recognition, former professor at Johns Hopkins University, and the creator of the open-source speech recognition tool Kaldi, announced on Twitter that he will be joining Xiaomi at the end of 2019. Currently, he is in the process of finalizing the contract. After joining Xiaomi, he will lead a team to develop the next generation of PyTorch-y Kaldi.
What prompted such a prominent figure in the AI field to choose Xiaomi, a company that is not considered top-tier in the field of artificial intelligence?
Who is Daniel Povey?
Daniel Povey, a British national, holds a Ph.D. from the University of Cambridge and was an assistant professor at Johns Hopkins University. He is an expert in the field of speech recognition and has led the development of Kaldi, a recognition toolkit that supports various speech recognition models. Many domestic and international speech technology companies have started their research and development testing with Kaldi.
Kaldi aims to provide flexible and scalable software for speech recognition developers. It supports linear transforms, MMI, boosted MMI, and MCE discriminative training, feature space discriminative training, and deep neural networks. It has also integrated with TensorFlow. As one of the most widely used open-source speech recognition toolkits, Kaldi has had a significant impact.
With over 6600+ stars on GitHub, Kaldi can be considered the cornerstone of the industry's speech recognition frameworks. It integrates various speech recognition models, including hidden Markov models and the latest deep learning neural networks, and holds an unshakable position in the field of speech recognition.
In addition, Dan Povey is also one of the authors of the HTK (Hidden Markov Model Toolkit) developed by Professor Phil Woodland, who led the development of the HTK toolkit at the University of Cambridge. After several years of development, Kaldi has integrated more technologies than HTK, making it the second-generation open-source speech recognition software developed by Professor Woodland's student. This is a widely circulated story in the industry.
Povey is not only the main developer of the speech recognition software Kaldi but also the first author of many papers. Since 1999, he has been publishing research results in the field of speech recognition and processing. To date, his research has been cited over 20,000 times on Google Scholar. Therefore, it is not an exaggeration to say that Povey holds a significant position in the field of AI speech recognition.
The Cause - "A Farce"
As a prominent figure in the AI field, Povey is undoubtedly a highly sought-after talent by major technology companies. So why did he choose Xiaomi, which is relatively "ordinary," over top-tier companies like Facebook?
It all started with a farce in August.
Povey's previous job was an assistant research professor at the Center for Language and Speech Processing at Johns Hopkins University, a position he held for seven years. However, in August of this year, a student protest led to his dismissal from Johns Hopkins University.
It is understood that in August, students at Johns Hopkins University protested against the campus police department's training contract with the U.S. Immigration and Customs Enforcement, fearing that it would lead to violence against students. The protest took the form of closing doors and windows, sitting in front of the Garland Hall, where the CLSP (Center for Language and Speech Processing) servers storing core data were located. These servers were very delicate, and any problems would require immediate on-site maintenance to avoid affecting research progress.
In order to maintain the servers, Povey organized a team of "counter-protesters" and attempted to retrieve the servers from the students. Physical conflicts inevitably occurred during this struggle. Subsequently, Povey was accused by students and related organizations of being a "racist." Under pressure, the university ultimately dismissed Povey, citing concerns for student safety.
Extension - Facebook's "Olive Branch"
After being dismissed by the university, Povey expressed strong dissatisfaction on his blog, as he felt he had been treated with double standards.
If the story had ended there, there might not have been any involvement with Xiaomi. However, reality is often "interesting." After the protest, Povey received an offer from Facebook with a salary of several million dollars per year. Feeling unfairly treated, Povey immediately accepted the offer and announced it on his personal blog.
For Povey, who was filled with anger, Facebook's offer was like a refreshing glass of iced cola on a hot summer day. Unfortunately, the glass contained not cola but oil.
According to Povey, on August 15th, Facebook sent him an email stating that they might not be able to hire him as a full-time employee due to the independent investigation into his dismissal from JHU. Unable to tolerate this, Povey, who was already filled with anger, immediately issued a statement:
"I refuse Facebook and will turn to Chinese companies or universities."
This news immediately sparked widespread discussion within the industry, with many speculating about which company Povey would choose. It was not until yesterday that Povey announced his decision to join Xiaomi, which was confirmed by insiders at Xiaomi, stating that the final procedures are currently underway.
Why Choose Xiaomi?
There are numerous Chinese companies, so why did Povey choose Xiaomi?
Speculation suggests that the reason lies in Xiaomi's ambition and strategy in the field of speech recognition, as well as its open-source development approach, which aligns with Povey's open-source philosophy.
In addition, it is likely due to Xiaomi's sincerity. It is reported that Cui Baoqiu, Vice President of Xiaomi Group and Chairman of the Technical Committee, personally welcomed Povey to China, showing great sincerity.
In any case, Xiaomi has won the talent war, snatching a prominent figure in the field of AI speech recognition from competitors such as Alibaba and Tencent. This undoubtedly serves as a boost for Xiaomi, which has been dedicated to the development of "Xiao Ai" (Xiaomi's virtual assistant).
Finally, we wish Mr. Daniel Povey a pleasant work experience in China.
(Reference sources: AI Technology and Industry Community, Sina Finance, Titanium Media, Zhai Shu Digital Kael, Phoenix Technology, AI Technology Headquarters, Google News)
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.