iFlytek releases Spark-ASR-2.0 speech recognition large model
According to IT Home, iFlytek released its latest-generation speech recognition large model, Spark-ASR-2.0, on September 23. The new model features key technological innovations and improvements in areas such as general recognition and complex acoustic scenarios, while its overall inference cost increased by only 10% compared to Spark-ASR-1.0.
Key facts
- 01Spark-ASR-2.0 integrates non-autoregressive and LLM-enhanced autoregressive synergy, mixed Chinese-English text and acoustic joint enhancement, and dynamic context injection.
- 02The model improves performance in general recognition including mixed Chinese-English, dialects, and terminology, as well as complex acoustic scenarios like high noise, low volume, fast speech, and children's speech.
- 03Compared to Spark-ASR-1.0, the new model significantly reduces the word error rate across core speech recognition scenarios and exhibits lower WER and CER in test sets.
- 04The official statement indicates that Spark-ASR-2.0 holds significant advantages in dialects, high noise, and low volume compared to current industry-best levels, outperforming industry optimums on major tasks.
AI-generated from the sources below. Always check the originals.
How each country tells it
So far our sources show coverage from one country. Perspectives appear once we find reports from a second country.
Sources
Summaries are AI-generated from the linked sources and may contain errors; always check the originals. We summarise and link; we never republish articles. Photos come from openly licensed libraries, official publicity material and brand logos, credited to their sources. If you own an image and want it credited differently or removed, email info@coda.news and we will act promptly.