我注意到使用Apple Speech Framework的一种奇怪行为。
在我自己的应用程序中实现并使用Apple和在线提供的示例后,我注意到语音框架总是在语音检测结果后为每个转录段返回相同的置信度。
以下是我在本Q底部使用原始转储执行的测试的格式化数据,您会注意到,对于每个SFTranscription,索引的段具有相同的置信度值,尽管具有不同转录的不同对象。
我认为这是不正确的,因为它在文档中说,结果是自信地排序,首先是最自信的。
如果有人知道对此有所修正,或者可以将我指向明确承认存在问题的地方,那将是最有帮助的。
[
<SFTranscription: 0x1c0032320>, formattedString=Hey there what's going on with you today how are you looking for things are you alright,
segments=(
<SFTranscriptionSegment: 0x1c0275340> confidence=0.931,
<SFTranscriptionSegment: 0x1c0275000> confidence=0.928,
<SFTranscriptionSegment: 0x1c0270e00> confidence=0.928,
<SFTranscriptionSegment: 0x1c0275dc0> confidence=0.924,
<SFTranscriptionSegment: 0x1c0275040> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275300> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275b00> confidence=0.925,
<SFTranscriptionSegment: 0x1c026db80> confidence=0.925,
<SFTranscriptionSegment: 0x1c026dbc0> confidence=0.925,
<SFTranscriptionSegment: 0x1c0271380> confidence=0.927,
<SFTranscriptionSegment: 0x1c026d9c0> confidence=0.818349,
<SFTranscriptionSegment: 0x1c026db00> confidence=0.942,
<SFTranscriptionSegment: 0x1c0271b00> confidence=0.826,
),
<SFTranscription: 0x1c002fba0>, formattedString=Hello what's going on with you today how are you looking for things are you alright,
segments=(
<SFTranscriptionSegment: 0x1c026dc80> confidence=0.6732572,
<SFTranscriptionSegment: 0x1c02752c0> confidence=0.931,
<SFTranscriptionSegment: 0x1c0275600> confidence=0.928,
<SFTranscriptionSegment: 0x1c0276b80> confidence=0.924,
<SFTranscriptionSegment: 0x1c00760c0> confidence=0.925,
<SFTranscriptionSegment: 0x1c0075c40> confidence=0.925,
<SFTranscriptionSegment: 0x1c0075e00> confidence=0.925,
<SFTranscriptionSegment: 0x1c00719c0> confidence=0.925,
<SFTranscriptionSegment: 0x1c0075740> confidence=0.925,
<SFTranscriptionSegment: 0x1c0076f00> confidence=0.927,
<SFTranscriptionSegment: 0x1c0276980> confidence=0.818349,
<SFTranscriptionSegment: 0x1c0279200> confidence=0.942
<SFTranscriptionSegment: 0x1c027bcc0> confidence=0.826,
),
<SFTranscription: 0x1c0030c00>, formattedString=Hey there what's going on with you today how are you looking for things that you alright,
segments=(
<SFTranscriptionSegment: 0x1c0275340> confidence=0.6732572,
<SFTranscriptionSegment: 0x1c027bd00> confidence=0.931,
<SFTranscriptionSegment: 0x1c0275000> confidence=0.928,
<SFTranscriptionSegment: 0x1c0270e00> confidence=0.928,
<SFTranscriptionSegment: 0x1c0275dc0> confidence=0.924,
<SFTranscriptionSegment: 0x1c0275040> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275300> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275b00> confidence=0.925,
<SFTranscriptionSegment: 0x1c026db80> confidence=0.925,
<SFTranscriptionSegment: 0x1c026dbc0> confidence=0.925,
<SFTranscriptionSegment: 0x1c0271380> confidence=0.927,
<SFTranscriptionSegment: 0x1c027bc80> confidence=0.818349,
<SFTranscriptionSegment: 0x1c027bd40> confidence=0.942,
<SFTranscriptionSegment: 0x1c027bd80> confidence=0.826,
),
<SFTranscription: 0x1c0033860>, formattedString=Hey there what's going on with you today how are you looking for things are you are right, segments=(
<SFTranscriptionSegment: 0x1c0275340> confidence=0.6732572,
<SFTranscriptionSegment: 0x1c027bd00> confidence=0.931,
<SFTranscriptionSegment: 0x1c0275000> confidence=0.928,
<SFTranscriptionSegment: 0x1c0270e00> confidence=0.928,
<SFTranscriptionSegment: 0x1c0275dc0> confidence=0.924,
<SFTranscriptionSegment: 0x1c0275040> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275300> confidence=0.925,
<SFTranscriptionSegment: 0x1c0275b00> confidence=0.925,
<SFTranscriptionSegment: 0x1c026db80> confidence=0.925,
<SFTranscriptionSegment: 0x1c026dbc0> confidence=0.925,
<SFTranscriptionSegment: 0x1c0271380> confidence=0.927,
<SFTranscriptionSegment: 0x1c026d9c0> confidence=0.818349,
<SFTranscriptionSegment: 0x1c026db00> confidence=0.942,
<SFTranscriptionSegment: 0x1c027bdc0> confidence=0.826,
)]
[<SFTranscription: 0x1c0032320>, formattedString=Hey there what's going on with you today how are you looking for things are you alright, segments=(
"<SFTranscriptionSegment: 0x1c0275340>, substringRange={0, 16}, timestamp=0.28, duration=0.52, confidence=0.6732572, substring=Hey there what's, alternativeSubstrings=(\n \"Hello what's\"\n)",
"<SFTranscriptionSegment: 0x1c027bd00>, substringRange={17, 5}, timestamp=0.8, duration=0.26, confidence=0.931, substring=going, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275000>, substringRange={23, 2}, timestamp=1.06, duration=0.09999999999999987, confidence=0.928, substring=on, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0270e00>, substringRange={26, 4}, timestamp=1.16, duration=0.1500000000000001, confidence=0.928, substring=with, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275dc0>, substringRange={31, 3}, timestamp=1.31, duration=0.1099999999999999, confidence=0.924, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275040>, substringRange={35, 5}, timestamp=1.42, duration=0.3200000000000001, confidence=0.925, substring=today, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275300>, substringRange={41, 3}, timestamp=1.74, duration=0.1600000000000001, confidence=0.925, substring=how, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275b00>, substringRange={45, 3}, timestamp=1.9, duration=0.1100000000000001, confidence=0.925, substring=are, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026db80>, substringRange={49, 3}, timestamp=2.01, duration=0.1699999999999999, confidence=0.925, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026dbc0>, substringRange={53, 7}, timestamp=2.18, duration=0.3300000000000001, confidence=0.925, substring=looking, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0271380>, substringRange={61, 3}, timestamp=2.51, duration=0.1199999999999997, confidence=0.927, substring=for, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026d9c0>, substringRange={65, 10}, timestamp=2.63, duration=0.3800000000000003, confidence=0.818349, substring=things are, alternativeSubstrings=(\n \"things that\"\n)",
"<SFTranscriptionSegment: 0x1c026db00>, substringRange={76, 3}, timestamp=3.01, duration=0.1499999999999999, confidence=0.942, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0271b00>, substringRange={80, 7}, timestamp=3.16, duration=1.62, confidence=0.826, substring=alright, alternativeSubstrings=(\n \"are right\"\n)"
), <SFTranscription: 0x1c002fba0>, formattedString=Hello what's going on with you today how are you looking for things are you alright, segments=(
"<SFTranscriptionSegment: 0x1c026dc80>, substringRange={0, 12}, timestamp=0.28, duration=0.52, confidence=0.6732572, substring=Hello what's, alternativeSubstrings=(\n \"Hey there what's\"\n)",
"<SFTranscriptionSegment: 0x1c02752c0>, substringRange={13, 5}, timestamp=0.8, duration=0.26, confidence=0.931, substring=going, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0273500>, substringRange={19, 2}, timestamp=1.06, duration=0.09999999999999987, confidence=0.928, substring=on, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275600>, substringRange={22, 4}, timestamp=1.16, duration=0.1500000000000001, confidence=0.928, substring=with, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0276b80>, substringRange={27, 3}, timestamp=1.31, duration=0.1099999999999999, confidence=0.924, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c00760c0>, substringRange={31, 5}, timestamp=1.42, duration=0.3200000000000001, confidence=0.925, substring=today, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0075c40>, substringRange={37, 3}, timestamp=1.74, duration=0.1600000000000001, confidence=0.925, substring=how, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0075e00>, substringRange={41, 3}, timestamp=1.9, duration=0.1100000000000001, confidence=0.925, substring=are, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c00719c0>, substringRange={45, 3}, timestamp=2.01, duration=0.1699999999999999, confidence=0.925, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0075740>, substringRange={49, 7}, timestamp=2.18, duration=0.3300000000000001, confidence=0.925, substring=looking, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0076f00>, substringRange={57, 3}, timestamp=2.51, duration=0.1199999999999997, confidence=0.927, substring=for, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0276980>, substringRange={61, 10}, timestamp=2.63, duration=0.3800000000000003, confidence=0.818349, substring=things are, alternativeSubstrings=(\n \"things that\"\n)",
"<SFTranscriptionSegment: 0x1c0279200>, substringRange={72, 3}, timestamp=3.01, duration=0.1499999999999999, confidence=0.942, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c027bcc0>, substringRange={76, 7}, timestamp=3.16, duration=1.62, confidence=0.826, substring=alright, alternativeSubstrings=(\n \"are right\"\n)"
), <SFTranscription: 0x1c0030c00>, formattedString=Hey there what's going on with you today how are you looking for things that you alright, segments=(
"<SFTranscriptionSegment: 0x1c0275340>, substringRange={0, 16}, timestamp=0.28, duration=0.52, confidence=0.6732572, substring=Hey there what's, alternativeSubstrings=(\n \"Hello what's\"\n)",
"<SFTranscriptionSegment: 0x1c027bd00>, substringRange={17, 5}, timestamp=0.8, duration=0.26, confidence=0.931, substring=going, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275000>, substringRange={23, 2}, timestamp=1.06, duration=0.09999999999999987, confidence=0.928, substring=on, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0270e00>, substringRange={26, 4}, timestamp=1.16, duration=0.1500000000000001, confidence=0.928, substring=with, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275dc0>, substringRange={31, 3}, timestamp=1.31, duration=0.1099999999999999, confidence=0.924, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275040>, substringRange={35, 5}, timestamp=1.42, duration=0.3200000000000001, confidence=0.925, substring=today, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275300>, substringRange={41, 3}, timestamp=1.74, duration=0.1600000000000001, confidence=0.925, substring=how, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275b00>, substringRange={45, 3}, timestamp=1.9, duration=0.1100000000000001, confidence=0.925, substring=are, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026db80>, substringRange={49, 3}, timestamp=2.01, duration=0.1699999999999999, confidence=0.925, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026dbc0>, substringRange={53, 7}, timestamp=2.18, duration=0.3300000000000001, confidence=0.925, substring=looking, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0271380>, substringRange={61, 3}, timestamp=2.51, duration=0.1199999999999997, confidence=0.927, substring=for, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c027bc80>, substringRange={65, 11}, timestamp=2.63, duration=0.3800000000000003, confidence=0.818349, substring=things that, alternativeSubstrings=(\n \"things are\"\n)",
"<SFTranscriptionSegment: 0x1c027bd40>, substringRange={77, 3}, timestamp=3.01, duration=0.1499999999999999, confidence=0.942, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c027bd80>, substringRange={81, 7}, timestamp=3.16, duration=1.62, confidence=0.826, substring=alright, alternativeSubstrings=(\n \"are right\"\n)"
), <SFTranscription: 0x1c0033860>, formattedString=Hey there what's going on with you today how are you looking for things are you are right, segments=(
"<SFTranscriptionSegment: 0x1c0275340>, substringRange={0, 16}, timestamp=0.28, duration=0.52, confidence=0.6732572, substring=Hey there what's, alternativeSubstrings=(\n \"Hello what's\"\n)",
"<SFTranscriptionSegment: 0x1c027bd00>, substringRange={17, 5}, timestamp=0.8, duration=0.26, confidence=0.931, substring=going, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275000>, substringRange={23, 2}, timestamp=1.06, duration=0.09999999999999987, confidence=0.928, substring=on, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0270e00>, substringRange={26, 4}, timestamp=1.16, duration=0.1500000000000001, confidence=0.928, substring=with, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275dc0>, substringRange={31, 3}, timestamp=1.31, duration=0.1099999999999999, confidence=0.924, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275040>, substringRange={35, 5}, timestamp=1.42, duration=0.3200000000000001, confidence=0.925, substring=today, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275300>, substringRange={41, 3}, timestamp=1.74, duration=0.1600000000000001, confidence=0.925, substring=how, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0275b00>, substringRange={45, 3}, timestamp=1.9, duration=0.1100000000000001, confidence=0.925, substring=are, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026db80>, substringRange={49, 3}, timestamp=2.01, duration=0.1699999999999999, confidence=0.925, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026dbc0>, substringRange={53, 7}, timestamp=2.18, duration=0.3300000000000001, confidence=0.925, substring=looking, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c0271380>, substringRange={61, 3}, timestamp=2.51, duration=0.1199999999999997, confidence=0.927, substring=for, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c026d9c0>, substringRange={65, 10}, timestamp=2.63, duration=0.3800000000000003, confidence=0.818349, substring=things are, alternativeSubstrings=(\n \"things that\"\n)",
"<SFTranscriptionSegment: 0x1c026db00>, substringRange={76, 3}, timestamp=3.01, duration=0.1499999999999999, confidence=0.942, substring=you, alternativeSubstrings=(\n)",
"<SFTranscriptionSegment: 0x1c027bdc0>, substringRange={80, 9}, timestamp=3.16, duration=1.62, confidence=0.826, substring=are right, alternativeSubstrings=(\n alright\n)"
)]