云服务器特惠 99/年起
  1. 腾讯云优惠网首页
  2. 腾讯云产品

口语评测 评测维度

• 腾讯云产品

建议评分(SuggestedScore)，取值范围[0,100]，评分方式为建议评分 = 精准度（PronAccuracy）× 完整度（PronCompletion）×（2 – 完整度（PronCompletion）），如若评分策略不符合请参考 Words 数组中的详细分数自定义评分逻辑。

相关产品
口语评测（新版） 产品页，含版本规格、价格与购买入口。

一、建议评分

评测模式 建议评分
句子模式段落模式句子多分支模式 精准度（PronAccuracy）× 完整度（PronCompletion）×（2 – 完整度（PronCompletion））
单词评测模式单词纠错模式拼音模式 取精准度评分即可
自由说评测情景评测模式 可能需要根据返回结果进一步计算。如果无相关需求，自由说模式可以取精准度；情景评测模式当切题度（PronCompletion）为1时，取精准度。当切题度（PronCompletion）为-1时，评分为0
单词实时评测 取中间结果（SentenceInfo）的精准度

二、精准度

发音精准度（PronAccuracy），取值范围[0,100], 评判标准是在参照了标准发音以及语音库同种发音对比的前提下，综合参照元音字母以及组合在不同音节中的不同读法、辅音以及成音节的读音、单词重音、句子重音等不同的发音指标，给出发音精准度的得分。

总精准度是所有匹配单词的平均值。

三、流利度

发音流利度 (PronFluency), 取值范围[0-1.0], 是根据发音是否通顺给出基础进行打分，并考虑发音变化中连读、失去爆破、弱读、同化的表达，综合参照语调和节奏中意群与停顿、语调、节奏的发音情况，给出流利度的得分。

总流利度是所有匹配单词的平均值。

四、完整度

发音完整度 (PronCompletion), 取值范围[0-1.0], 是根据发音识别文本与上传 ref_text 文本的对比，是所有匹配单词占 ref_text 文本长度的比值。

五、匹配情况

MatchTag 表示当前音频数据的词和当前文本词的匹配情况。

0：匹配单词、1：新增单词、2：缺少单词、3：错读的词、4：未录入单词。

六、当前词与参考词

智聆口语评测进行评测的时候会对文本进行处理，参考 评估文本介绍。

处理后会按照当前词（Word）来进行评测，原始文本可以参考（ReferenceWord）。

七、请求参数

# 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&…eval_mode=1keyword=rec_mode=0ref_text=1st,wakey i go to school by busscore_coeff=1.000000sentence_info_enabled=0server_engine_type=16k_entext_mode=0voice_format=1voice_id=6bd18918-8d15-4272-b693-6c588f058a33

八、返回结果

{
"code": 0,
"message": "6bd18918-8d15-4272-b693-6c588f058a33_12",
"voice_id": "6bd18918-8d15-4272-b693-6c588f058a33",
"result": {
"SuggestedScore": 56.42676544189453,
"PronAccuracy": 61.44247817993164,
"PronFluency": 0.976434051990509,
"PronCompletion": 0.7142857313156128,
"Words": [
{
"MemBeginTime": 140,
"MemEndTime": 550,
"PronAccuracy": -1,
"PronFluency": 0,
"ReferenceWord": "*",
"Word": "*",
"MatchTag": 1,
"KeywordTag": 0,
"PhoneInfos": [
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 550,
"MemEndTime": 910,
"PronAccuracy": 99.0815658569336,
"PronFluency": 0.9839948415756226,
"ReferenceWord": "1st_0",
"Word": "first",
"MatchTag": 0,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 550,
"MemEndTime": 650,
"PronAccuracy": 98.95162963867188,
"DetectedStress": false,
"Phone": "f",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 650,
"MemEndTime": 770,
"PronAccuracy": 99.2125244140625,
"DetectedStress": false,
"Phone": "er",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 770,
"MemEndTime": 840,
"PronAccuracy": 99.16388702392578,
"DetectedStress": false,
"Phone": "s",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 840,
"MemEndTime": 910,
"PronAccuracy": 98.99822235107422,
"DetectedStress": false,
"Phone": "t",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 910,
"MemEndTime": 1210,
"PronAccuracy": 42.349395751953125,
"PronFluency": 0.9971372485160828,
"ReferenceWord": "wakey_2",
"Word": "wakey",
"MatchTag": 4,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 910,
"MemEndTime": 970,
"PronAccuracy": 98.7578353881836,
"DetectedStress": false,
"Phone": "w",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 970,
"MemEndTime": 1050,
"PronAccuracy": 98.17191314697266,
"DetectedStress": false,
"Phone": "ey",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1050,
"MemEndTime": 1130,
"PronAccuracy": 14.024602890014648,
"DetectedStress": false,
"Phone": "k",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1130,
"MemEndTime": 1180,
"PronAccuracy": 0.4065273702144623,
"DetectedStress": false,
"Phone": "w",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1180,
"MemEndTime": 1210,
"PronAccuracy": 0.38612863421440125,
"DetectedStress": false,
"Phone": "ay",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 1210,
"MemEndTime": 1310,
"PronAccuracy": 2.167337656021118,
"PronFluency": 0.9951702952384949,
"ReferenceWord": "i_3",
"Word": "i",
"MatchTag": 0,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 1210,
"MemEndTime": 1310,
"PronAccuracy": 2.167337656021118,
"DetectedStress": false,
"Phone": "ay",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 1310,
"MemEndTime": 1490,
"PronAccuracy": 0.26200056076049805,
"PronFluency": 0.9832581281661987,
"ReferenceWord": "go_4",
"Word": "go",
"MatchTag": 0,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 1310,
"MemEndTime": 1390,
"PronAccuracy": 0.5108872056007385,
"DetectedStress": false,
"Phone": "g",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1390,
"MemEndTime": 1490,
"PronAccuracy": 0.013113964349031448,
"DetectedStress": false,
"Phone": "ow",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 1490,
"MemEndTime": 1650,
"PronAccuracy": 49.854270935058594,
"PronFluency": 0.9747055768966675,
"ReferenceWord": "to_5",
"Word": "to",
"MatchTag": 0,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 1490,
"MemEndTime": 1560,
"PronAccuracy": 0.8929398655891418,
"DetectedStress": false,
"Phone": "t",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1560,
"MemEndTime": 1650,
"PronAccuracy": 98.81560516357422,
"DetectedStress": false,
"Phone": "ah",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 1650,
"MemEndTime": 2250,
"PronAccuracy": 98.87284851074219,
"PronFluency": 0.924338161945343,
"ReferenceWord": "school_6",
"Word": "school",
"MatchTag": 0,
"KeywordTag": 0,
"PhoneInfos": [
{
"MemBeginTime": 1650,
"MemEndTime": 1750,
"PronAccuracy": 98.44365692138672,
"DetectedStress": false,
"Phone": "s",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1750,
"MemEndTime": 1830,
"PronAccuracy": 99.21931457519531,
"DetectedStress": false,
"Phone": "k",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1830,
"MemEndTime": 1990,
"PronAccuracy": 99.09530639648438,
"DetectedStress": false,
"Phone": "uw",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
},
{
"MemBeginTime": 1990,
"MemEndTime": 2250,
"PronAccuracy": 98.73311614990234,
"DetectedStress": false,
"Phone": "l",
"ReferencePhone": "",
"ReferenceLetter": "",
"Stress": false,
"MatchTag": 0
}
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 2250,
"MemEndTime": 2250,
"PronAccuracy": -1,
"PronFluency": 0,
"ReferenceWord": "by_7",
"Word": "by",
"MatchTag": 2,
"KeywordTag": 0,
"PhoneInfos": [
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
},
{
"MemBeginTime": 2250,
"MemEndTime": 2250,
"PronAccuracy": -1,
"PronFluency": 0,
"ReferenceWord": "bus_8",
"Word": "bus",
"MatchTag": 2,
"KeywordTag": 0,
"PhoneInfos": [
],
"Tone": {
"Valid": false,
"RefTone": -1,
"HypothesisTone": -1
}
}
],
"SentenceId": -1,
"RefTextId": -1,
"KeyWordHits": null,
"UnKeyWordHits": null
},
"final": 1
}

九、返回结果解析

音频发音内容为 hello first wakey i want to school。

·hello：由于评估文本中未存在这个单词，所以不会进行评测，此时 MatchTag 为1，Word为*。

·first：对应原文本 1st，此时 MatchTag 为0.评估的当前词 Word 为 first，评估文本参考词 ReferenceWord 为 1st_0。表示 1st 的下标为 0。

·wakey：此时 MatchTag 为4，由于该单词不属于常见词，所以我们暂未录入。一般情况下会尝试进行发音评测，如果不符合可以通过 音素标注。此时 Word 为 wakey，ReferenceWord 为 wakey_2。一般情况下标点符号不参与评测，但是会占位，所以此时表示wakey的下标为2，而不是1。

·i want to school：与评估文本 i go to school 有出入，但是此时 MatchTag 为0。评分偏低情况下，可以认为是错读的情况。

·by bus：此时 MatchTag 为2，由于未读该内容所以缺少单词。

总精准度只会计算当前词 MatchTag 为0的精准度的平均值。

完整度也只会计算当前词 MatchTag 为0的完整度。评估文本中虽然有8个单词，但是由于 wakey 是不在字典里面，MatchTag 为4。所以实际只有7个单词，MatchTag 为0的有5个单词，所以最终完整度为5/7。

相关内容：

  1. 口语评测 句子多分支评测模式（中文评测模式） # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_zheval_mode=6ref_text="我有一个苹果 | 我有一个香蕉 | 我有一个梨"score_coeff=1.000000voice_format=1。...
  2. 口语评测 句子评测模式（中文评测模式） # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_zheval_mode=1ref_text="我是谁"score_coeff=1.000000voice_format=1。...
  3. 口语评测 句子多分支评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_eneval_mode=6# 长音频推荐流式rec_mode=0# 流式展示中间结果sentence_info_enabled=1ref_text="i go to school by。...
  4. 口语评测 句子评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_eneval_mode=1# 流式rec_mode=0# 流式展示中间结果sentence_info_enabled=1ref_text="i go to school by bus"。...
  5. 口语评测 拼音评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_zheval_mode=8ref_text="shan3"score_coeff=1.000000voice_format=1。...
  6. 口语评测 单词纠错评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_eneval_mode=4rec_mode=1ref_text="happy"score_coeff=1.000000voice_format=1。...
  7. 口语评测 自由说评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_eneval_mode=3# 流式rec_mode=0# 流式展示中间结果sentence_info_enabled=1ref_text=""score_coeff=1.000000。...
  8. 口语评测 单词评测模式 # 参数示例为websocket连接URL展开, 如：soe.cloud.tencent.com/soe/api/1306***?eval_mode=0&voice_format=1&...server_engine_type=16k_eneval_mode=0rec_mode=0ref_text="father"score_coeff=1.000000sentence_info_enabled=0text_mod。...
赞 (0)
上一篇 4天前
下一篇 4天前