Hakha Chin STT v0.1.0

Hakha Chin STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Hakha Chin / Hakha Lai / cnh
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{hakhachin-stt, author = {Tyers,Francis}, title = {Hakha Chin STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-CNH-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Hakha Chin Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	77.8%	32.1%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Greek STT v0.1.0

Greek STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Greek / Ελληνικά / el
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{greek-stt, author = {Tyers,Francis}, title = {Greek STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-EL-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Greek Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	88.1%	36.3%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Georgian STT v0.1.0

Georgian STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Georgian / ქართული ენა / ka
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{georgian-stt, author = {Tyers,Francis}, title = {Georgian STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-KA-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Georgian Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	98.1%	34.7%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Frisian STT v0.1.0

Frisian STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Frisian / Frysk / fy-NL
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{frisian-stt, author = {Tyers,Francis}, title = {Frisian STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-FY_NL-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Chuvash Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	79.6.%	29.9%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Finnish STT v0.1.0

Finnish STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Finnish / Suomi / fi
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{finnish-stt, author = {Tyers,Francis}, title = {Finnish STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-FI-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Finnish Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	99.7%	39.1%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Estonian STT v0.1.0

Estonian STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Estonian / Eesti / et
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{estonian-stt, author = {Tyers,Francis}, title = {Estonian STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-ET-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Estonian Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	92.2%	29.5%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Dhivehi STT v0.1.0

Dhivehi STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Dhivehi / ދިވެހި / dv
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{dhivehi-stt, author = {Tyers,Francis}, title = {Dhivehi STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-DV-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Dhivehi Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	94.7%	33.0%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Chuvash STT v0.1.0

Chuvash STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Chuvash / Чӑвашла / cv
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{chuvash-stt, author = {Tyers,Francis}, title = {Chuvash STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-CV-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Chuvash Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	97.0%	36.9%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Breton STT v0.1.0

Breton STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Breton / Brezhoneg / br
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{breton-stt, author = {Tyers,Francis}, title = {Breton STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-BR-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Breton Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	94.9%	41.6%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.

Basque STT v0.1.0

Basque STT v0.1.0 (ITML)

Jump to section:

• Model details
• Intended use
• Performance Factors
• Metrics
• Training data
• Evaluation data
• Ethical considerations
• Caveats and recommendations

Model details

• Person or organization developing model: Originally trained by Francis Tyers and the Inclusive Technology for Marginalised Languages group.
• Model language: Basque / Euskara / eu
• Model date: April 9, 2021
• Model type: Speech-to-Text
• Model version: v0.1.0
• Compatible with 🐸 STT version: v0.9.3
• License: AGPL
• Citation details: @techreport{basque-stt, author = {Tyers,Francis}, title = {Basque STT 0.1}, institution = {Coqui}, address = {\url{https://github.com/coqui-ai/STT-models}} year = {2021}, month = {April}, number = {STT-CV6.1-EU-0.1} }
• Where to send questions or comments about the model: You can leave an issue on STT-model issues, open a new discussion on STT-model discussions, or chat with us on Gitter.

Intended use

Speech-to-Text for the Basque Language on 16kHz, mono-channel audio.

Performance Factors

Factors relevant to Speech-to-Text performance include but are not limited to speaker demographics, recording quality, and background noise. Read more about STT performance factors here.

Metrics

STT models are usually evaluated in terms of their transcription accuracy, deployment Real-Time Factor, and model size on disk.

Transcription Accuracy

The following Word Error Rates and Character Error Rates are reported on omnilingo.

Test Corpus	WER	CER
Common Voice	81.0%	19.9%

Real-Time Factor

Real-Time Factor (RTF) is defined as proccesing-time / length-of-audio. The exact real-time factor of an STT model will depend on the hardware setup, so you may experience a different RTF.

Recorded average RTF on laptop CPU: ``

Model Size

model.pbmm: 181M
model.tflite: 46M

Approaches to uncertainty and variability

Confidence scores and multiple paths from the decoding beam can be used to measure model uncertainty and provide multiple, variable transcripts for any processed audio.

Training data

This model was trained on Common Voice 6.1 train.

Evaluation data

The Model was evaluated on Common Voice 6.1 test.

Ethical considerations

Deploying a Speech-to-Text model into any production setting has ethical implications. You should consider these implications before use.

Demographic Bias

You should assume every machine learning model has demographic bias unless proven otherwise. For STT models, it is often the case that transcription accuracy is better for men than it is for women. If you are using this model in production, you should acknowledge this as a potential issue.

Surveillance

Speech-to-Text may be mis-used to invade the privacy of others by recording and mining information from private conversations. This kind of individual privacy is protected by law in may countries. You should not assume consent to record and analyze private speech.

Caveats and recommendations

Machine learning models (like this STT model) perform best on data that is similar to the data on which they were trained. Read about what to expect from an STT model with regard to your data here.

In most applications, it is recommended that you train your own language model to improve transcription accuracy on your speech data.