跳到主内容
@wquguru
精选86Rohan Paul模型发布/更新

WisprFlow发布自研语音模型Canto 1,WER仅3.4%

The keyboard was a workaround for machines that couldn’t hear. Machines that hea…

原文
发到 X
推荐理由

自研语音模型在真实嘈杂场景下实现极低词错率,且详细披露了基于用户反馈的RL训练闭环,对做语音Agent的同学极具参考价值。

The keyboard was a workaround for machines that couldn’t hear. Machines that hear make it optional.

键盘是为那些无法“听”的机器设计的权宜之计。能够“听”的机器让键盘变得可有可无。

WisprFlow recently hired Ariya Rastrow (founding Alexa scientist) as chief science officer and moved from rented speech engines to models it trains itself.

WisprFlow 最近聘请了 Alexa 的创始科学家 Ariya Rastrow 担任首席科学官,并从使用租赁的语音引擎转向训练自己的模型。

Hard to find a better fit for a speech lab than someone who scaled Alexa’s speech team.

很难找到比曾将 Alexa 语音团队规模化的人更合适的人选来领导语音实验室。

Canto 1, his lab’s first model, was built to cut how often people edit dictated text, so Wispr trained it on noisy real-world dictations and tuned it to each user’s vocabulary and names.

Canto 1 是其实验室的第一个模型,旨在减少人们编辑语音输入文本的频率,因此 Wispr 在嘈杂的真实世界语音输入数据上对其进行训练,并针对每个用户的词汇和姓名进行了调整。

And this is a hard test, unlike many, because Speech recognition was long tested on clean, read-aloud audio, but the hard problem is a cheap laptop mic in a room where someone else is talking.

这是一个艰难的测试,与许多其他测试不同,因为语音识别长期以来一直是在干净、朗读式的音频上进行测试的,但真正的难题是在有人说话的房间中使用廉价的笔记本电脑麦克风进行识别。

On about 10 hours of real Flow dictations from more than 2,300 speakers, it posted a 3.4% word error rate, lower than all competing models that Wispr tested.

在来自 2,300 多名发言者的约 10 小时真实 Flow 语音输入数据上,其词错误率(WER)为 3.4%,低于 Wispr 测试的所有竞争模型。

Typing gives you feedback on every keystroke, while speech makes you wait until the end of a sentence to see what the machine heard, so speed counts as much as accuracy.

打字会在每次按键时提供反馈,而语音则让你必须等到句子结束才能看到机器听到了什么,因此速度的重要性不亚于准确性。

Since launch, dictations that take more than 1 second to appear are down 54%.

自发布以来,出现时间超过 1 秒的语音输入减少了 54%。

Its training runs through pretraining, supervised fine-tuning and reinforcement learning with GRPO, the same stages used for large language models.

其训练过程包括预训练、监督微调和基于 GRPO 的强化学习,这与大型语言模型所使用的阶段相同。

The reinforcement learning stage uses corrections from Flow users who opted in to share data.

强化学习阶段使用了选择共享数据的 Flow 用户的纠正内容。

When someone fixes a dictated word, Wispr checks how the audio aligns with the text and how the edit is shaped.

当某人修正一个语音输入的单词时,Wispr 会检查音频如何与文本对齐,以及编辑是如何形成的。

A misheard word gets folded back into the training reference, and a plain rewrite of the sentence gets ignored.

被误听的单词会被重新纳入训练参考中,而单纯的句子重写则会被忽略。

更进一步:量化金融体系

看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力

进入量化体系 →

相似阅读

关联信息,但可能不是同一事件