Blame - docs/dyn/speech_v1.speech.html - platform/external/python/google-api-python-client

2017-06-06 18:46:08 -0400

[diff] [blame]

115

# process the request.

116

# request.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

117

"speechContexts": [ # Array of SpeechContext.

118

# A means to provide context to assist the speech recognition. For more

119

# information, see

120

# [speech

121

# adaptation](https://cloud.google.com/speech-to-text/docs/context-strength).

122

{ # Provides "hints" to the speech recognizer to favor specific words and phrases

123

# in the results.

124

"phrases": [ # A list of strings containing words and phrases "hints" so that

125

# the speech recognition is more likely to recognize them. This can be used

126

# to improve the accuracy for specific words and phrases, for example, if

127

# specific commands are typically spoken by the user. This can also be used

128

# to add additional words to the vocabulary of the recognizer. See

129

# [usage limits](https://cloud.google.com/speech-to-text/quotas#content).

130

#

131

# List items can also be set to classes for groups of words that represent

132

# common concepts that occur in natural language. For example, rather than

133

# providing phrase hints for every month of the year, using the $MONTH class

134

# improves the likelihood of correctly transcribing audio that includes

# months.

"A String",

],

},

],

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

140

"languageCode": "A String", # Required. The language of the supplied audio as a

141

# [BCP-47](https://www.rfc-editor.org/rfc/bcp/bcp47.txt) language tag.

142

# Example: "en-US".

143

# See [Language

144

# Support](https://cloud.google.com/speech-to-text/docs/languages) for a list

145

# of the currently supported language codes.

146

"useEnhanced": True or False, # Set to true to use an enhanced model for speech recognition.

147

# If `use_enhanced` is set to true and the `model` field is not set, then

148

# an appropriate enhanced model is chosen if an enhanced model exists for

149

# the audio.

150

#

151

# If `use_enhanced` is true and an enhanced version of the specified model

152

# does not exist, then the speech is recognized using the standard version

153

# of the specified model.

154

"enableWordTimeOffsets": True or False, # If `true`, the top result includes a list of words and

155

# the start and end time offsets (timestamps) for those words. If

156

# `false`, no word-level time offset information is returned. The default is

157

# `false`.

158

"diarizationConfig": { # Config to enable speaker diarization. # Config to enable speaker diarization and set additional

159

# parameters to make diarization better suited for your application.

160

# Note: When this is enabled, we send all the words from the beginning of the

161

# audio for the top alternative in every consecutive STREAMING responses.

162

# This is done in order to improve our speaker tags as our models learn to

163

# identify the speakers in the conversation over time.

164

# For non-streaming requests, the diarization results will be provided only

165

# in the top alternative of the FINAL SpeechRecognitionResult.

166

"speakerTag": 42, # Output only. Unused.

167

"minSpeakerCount": 42, # Minimum number of speakers in the conversation. This range gives you more

168

# flexibility by allowing the system to automatically determine the correct

169

# number of speakers. If not set, the default value is 2.

170

"enableSpeakerDiarization": True or False, # If 'true', enables speaker detection for each recognized word in

171

# the top alternative of the recognition result using a speaker_tag provided

172

# in the WordInfo.

173

"maxSpeakerCount": 42, # Maximum number of speakers in the conversation. This range gives you more

174

# flexibility by allowing the system to automatically determine the correct

175

# number of speakers. If not set, the default value is 6.

176

},

177

"profanityFilter": True or False, # If set to `true`, the server will attempt to filter out

178

# profanities, replacing all but the initial character in each filtered word

179

# with asterisks, e.g. "f***". If set to `false` or omitted, profanities

180

# won't be filtered out.

181

"enableAutomaticPunctuation": True or False, # If 'true', adds punctuation to recognition result hypotheses.

182

# This feature is only available in select languages. Setting this for

183

# requests in other languages has no effect at all.

184

# The default 'false' value does not add punctuation to result hypotheses.

185

"enableSeparateRecognitionPerChannel": True or False, # This needs to be set to `true` explicitly and `audio_channel_count` > 1

186

# to get each channel recognized separately. The recognition result will

187

# contain a `channel_tag` field to state which channel that result belongs

188

# to. If this is not true, we will only recognize the first channel. The

189

# request is billed cumulatively for all channels recognized:

190

# `audio_channel_count` multiplied by the length of the audio.

191

"sampleRateHertz": 42, # Sample rate in Hertz of the audio data sent in all

192

# `RecognitionAudio` messages. Valid values are: 8000-48000.

193

# 16000 is optimal. For best results, set the sampling rate of the audio

194

# source to 16000 Hz. If that's not possible, use the native sample rate of

195

# the audio source (instead of re-sampling).

196

# This field is optional for FLAC and WAV audio files, but is

197

# required for all other audio formats. For details, see AudioEncoding.

198

"metadata": { # Description of audio data to be recognized. # Metadata regarding this request.

199

"originalMediaType": "A String", # The original media the speech was recorded on.

200

"recordingDeviceName": "A String", # The device used to make the recording. Examples 'Nexus 5X' or

201

# 'Polycom SoundStation IP 6000' or 'POTS' or 'VoIP' or

202

# 'Cardioid Microphone'.

203

"industryNaicsCodeOfAudio": 42, # The industry vertical to which this speech recognition request most

204

# closely applies. This is most indicative of the topics contained

205

# in the audio. Use the 6-digit NAICS code to identify the industry

206

# vertical - see https://www.naics.com/search/.

207

"originalMimeType": "A String", # Mime type of the original audio file. For example `audio/m4a`,

208

# `audio/x-alaw-basic`, `audio/mp3`, `audio/3gpp`.

209

# A list of possible audio mime types is maintained at

210

# http://www.iana.org/assignments/media-types/media-types.xhtml#audio

211

"audioTopic": "A String", # Description of the content. Eg. "Recordings of federal supreme court

212

# hearings from 2012".

213

"recordingDeviceType": "A String", # The type of device the speech was recorded with.

214

"microphoneDistance": "A String", # The audio type that most closely describes the audio being recognized.

215

"interactionType": "A String", # The use case most closely describing the audio content to be recognized.

216

},

217

"maxAlternatives": 42, # Maximum number of recognition hypotheses to be returned.

218

# Specifically, the maximum number of `SpeechRecognitionAlternative` messages

219

# within each `SpeechRecognitionResult`.

220

# The server may return fewer than `max_alternatives`.

221

# Valid values are `0`-`30`. A value of `0` or `1` will return a maximum of

222

# one. If omitted, will return a maximum of one.

223

"audioChannelCount": 42, # The number of channels in the input audio data.

224

# ONLY set this for MULTI-CHANNEL recognition.

225

# Valid values for LINEAR16 and FLAC are `1`-`8`.

226

# Valid values for OGG_OPUS are '1'-'254'.

227

# Valid value for MULAW, AMR, AMR_WB and SPEEX_WITH_HEADER_BYTE is only `1`.

228

# If `0` or omitted, defaults to one channel (mono).

229

# Note: We only recognize the first channel by default.

230

# To perform independent recognition on each channel set

231

# `enable_separate_recognition_per_channel` to 'true'.

Bu Sun Kim

2020-05-27 12:20:54 -0700

[diff] [blame]

232

"encoding": "A String", # Encoding of audio data sent in all `RecognitionAudio` messages.

233

# This field is optional for `FLAC` and `WAV` audio files and required

234

# for all other audio formats. For details, see AudioEncoding.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

235

"model": "A String", # Which model to select for the given request. Select the model

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

236

# best suited to your domain to get best results. If a model is not

237

# explicitly specified, then we auto-select a model based on the parameters

238

# in the RecognitionConfig.

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

239

# <table>

240

# <tr>

241

# <td><b>Model</b></td>

242

# <td><b>Description</b></td>

243

# </tr>

244

# <tr>

245

# <td><code>command_and_search</code></td>

246

# <td>Best for short queries such as voice commands or voice search.</td>

247

# </tr>

248

# <tr>

249

# <td><code>phone_call</code></td>

250

# <td>Best for audio that originated from a phone call (typically

251

# recorded at an 8khz sampling rate).</td>

252

# </tr>

253

# <tr>

254

# <td><code>video</code></td>

255

# <td>Best for audio that originated from from video or includes multiple

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

256

# speakers. Ideally the audio is recorded at a 16khz or greater

257

# sampling rate. This is a premium model that costs more than the

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

258

# standard rate.</td>

259

# </tr>

260

# <tr>

261

# <td><code>default</code></td>

262

# <td>Best for audio that is not one of the specific audio models.

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

263

# For example, long-form audio. Ideally the audio is high-fidelity,

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

264

# recorded at a 16khz or greater sampling rate.</td>

265

# </tr>

266

# </table>

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

},

}

x__xgafv: string, V1 error format.

Allowed values

1 - v1 error format

2 - v2 error format

Returns:

An object of the form:

277

278

{ # This resource represents a long-running operation that is the result of a

279

# network API call.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

280

"response": { # The normal response of the operation in case of success. If the original

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

281

# method returns no data on success, such as `Delete`, the response is

282

# `google.protobuf.Empty`. If the original method is standard

283

# `Get`/`Create`/`Update`, the response should be the resource. For other

284

# methods, the response should have the type `XxxResponse`, where `Xxx`

285

# is the original method name. For example, if the original method name

286

# is `TakeSnapshot()`, the inferred response type is

287

# `TakeSnapshotResponse`.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

288

"a_key": "", # Properties of the object. Contains field @type with type URL.

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

289

},

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

290

"metadata": { # Service-specific metadata associated with the operation. It typically

291

# contains progress information and common metadata such as create time.

292

# Some services might not provide such metadata. Any method that returns a

293

# long-running operation should document the metadata type, if any.

294

"a_key": "", # Properties of the object. Contains field @type with type URL.

295

},

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

296

"name": "A String", # The server-assigned name, which is only unique within the same service that

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

297

# originally returns it. If you use the default HTTP mapping, the

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

298

# `name` should be a resource name ending with `operations/{unique_id}`.

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

299

"done": True or False, # If the value is `false`, it means the operation is still in progress.

300

# If `true`, the operation is completed, and either `error` or `response` is

301

# available.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

302

"error": { # The `Status` type defines a logical error model that is suitable for # The error result of the operation in case of failure or cancellation.

303

# different programming environments, including REST APIs and RPC APIs. It is

304

# used by [gRPC](https://github.com/grpc). Each `Status` message contains

305

# three pieces of data: error code, error message, and error details.

306

#

307

# You can find out more about this error model and how to work with it in the

308

# [API Design Guide](https://cloud.google.com/apis/design/errors).

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

309

"message": "A String", # A developer-facing error message, which should be in English. Any

310

# user-facing error message should be localized and sent in the

311

# google.rpc.Status.details field, or localized by the client.

312

"code": 42, # The status code, which should be an enum value of google.rpc.Code.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

313

"details": [ # A list of messages that carry the error details. There is a common set of

314

# message types for APIs to use.

315

{

316

"a_key": "", # Properties of the object. Contains field @type with type URL.

317

},

318

],

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

319

},

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

}</pre>

</div>

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

324

<code class="details" id="recognize">recognize(body=None, x__xgafv=None)</code>

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

325

<pre>Performs synchronous speech recognition: receive results after all audio

326

has been sent and processed.

327

328

Args:

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

329

body: object, The request body.

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

330

The object takes the form of:

331

332

{ # The top-level message sent by the client for the `Recognize` method.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

333

"config": { # Provides information to the recognizer that specifies how to process the # Required. Provides information to the recognizer that specifies how to

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

334

# process the request.

335

# request.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

336

"speechContexts": [ # Array of SpeechContext.

337

# A means to provide context to assist the speech recognition. For more

338

# information, see

339

# [speech

340

# adaptation](https://cloud.google.com/speech-to-text/docs/context-strength).

341

{ # Provides "hints" to the speech recognizer to favor specific words and phrases

342

# in the results.

343

"phrases": [ # A list of strings containing words and phrases "hints" so that

344

# the speech recognition is more likely to recognize them. This can be used

345

# to improve the accuracy for specific words and phrases, for example, if

346

# specific commands are typically spoken by the user. This can also be used

347

# to add additional words to the vocabulary of the recognizer. See

348

# [usage limits](https://cloud.google.com/speech-to-text/quotas#content).

349

#

350

# List items can also be set to classes for groups of words that represent

351

# common concepts that occur in natural language. For example, rather than

352

# providing phrase hints for every month of the year, using the $MONTH class

353

# improves the likelihood of correctly transcribing audio that includes

# months.

"A String",

],

},

],

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

359

"languageCode": "A String", # Required. The language of the supplied audio as a

360

# [BCP-47](https://www.rfc-editor.org/rfc/bcp/bcp47.txt) language tag.

361

# Example: "en-US".

362

# See [Language

363

# Support](https://cloud.google.com/speech-to-text/docs/languages) for a list

364

# of the currently supported language codes.

365

"useEnhanced": True or False, # Set to true to use an enhanced model for speech recognition.

366

# If `use_enhanced` is set to true and the `model` field is not set, then

367

# an appropriate enhanced model is chosen if an enhanced model exists for

368

# the audio.

369

#

370

# If `use_enhanced` is true and an enhanced version of the specified model

371

# does not exist, then the speech is recognized using the standard version

372

# of the specified model.

373

"enableWordTimeOffsets": True or False, # If `true`, the top result includes a list of words and

374

# the start and end time offsets (timestamps) for those words. If

375

# `false`, no word-level time offset information is returned. The default is

376

# `false`.

377

"diarizationConfig": { # Config to enable speaker diarization. # Config to enable speaker diarization and set additional

378

# parameters to make diarization better suited for your application.

379

# Note: When this is enabled, we send all the words from the beginning of the

380

# audio for the top alternative in every consecutive STREAMING responses.

381

# This is done in order to improve our speaker tags as our models learn to

382

# identify the speakers in the conversation over time.

383

# For non-streaming requests, the diarization results will be provided only

384

# in the top alternative of the FINAL SpeechRecognitionResult.

385

"speakerTag": 42, # Output only. Unused.

386

"minSpeakerCount": 42, # Minimum number of speakers in the conversation. This range gives you more

387

# flexibility by allowing the system to automatically determine the correct

388

# number of speakers. If not set, the default value is 2.

389

"enableSpeakerDiarization": True or False, # If 'true', enables speaker detection for each recognized word in

390

# the top alternative of the recognition result using a speaker_tag provided

391

# in the WordInfo.

392

"maxSpeakerCount": 42, # Maximum number of speakers in the conversation. This range gives you more

393

# flexibility by allowing the system to automatically determine the correct

394

# number of speakers. If not set, the default value is 6.

395

},

396

"profanityFilter": True or False, # If set to `true`, the server will attempt to filter out

397

# profanities, replacing all but the initial character in each filtered word

398

# with asterisks, e.g. "f***". If set to `false` or omitted, profanities

399

# won't be filtered out.

400

"enableAutomaticPunctuation": True or False, # If 'true', adds punctuation to recognition result hypotheses.

401

# This feature is only available in select languages. Setting this for

402

# requests in other languages has no effect at all.

403

# The default 'false' value does not add punctuation to result hypotheses.

404

"enableSeparateRecognitionPerChannel": True or False, # This needs to be set to `true` explicitly and `audio_channel_count` > 1

405

# to get each channel recognized separately. The recognition result will

406

# contain a `channel_tag` field to state which channel that result belongs

407

# to. If this is not true, we will only recognize the first channel. The

408

# request is billed cumulatively for all channels recognized:

409

# `audio_channel_count` multiplied by the length of the audio.

410

"sampleRateHertz": 42, # Sample rate in Hertz of the audio data sent in all

411

# `RecognitionAudio` messages. Valid values are: 8000-48000.

412

# 16000 is optimal. For best results, set the sampling rate of the audio

413

# source to 16000 Hz. If that's not possible, use the native sample rate of

414

# the audio source (instead of re-sampling).

415

# This field is optional for FLAC and WAV audio files, but is

416

# required for all other audio formats. For details, see AudioEncoding.

417

"metadata": { # Description of audio data to be recognized. # Metadata regarding this request.

418

"originalMediaType": "A String", # The original media the speech was recorded on.

419

"recordingDeviceName": "A String", # The device used to make the recording. Examples 'Nexus 5X' or

420

# 'Polycom SoundStation IP 6000' or 'POTS' or 'VoIP' or

421

# 'Cardioid Microphone'.

422

"industryNaicsCodeOfAudio": 42, # The industry vertical to which this speech recognition request most

423

# closely applies. This is most indicative of the topics contained

424

# in the audio. Use the 6-digit NAICS code to identify the industry

425

# vertical - see https://www.naics.com/search/.

426

"originalMimeType": "A String", # Mime type of the original audio file. For example `audio/m4a`,

427

# `audio/x-alaw-basic`, `audio/mp3`, `audio/3gpp`.

428

# A list of possible audio mime types is maintained at

429

# http://www.iana.org/assignments/media-types/media-types.xhtml#audio

430

"audioTopic": "A String", # Description of the content. Eg. "Recordings of federal supreme court

431

# hearings from 2012".

432

"recordingDeviceType": "A String", # The type of device the speech was recorded with.

433

"microphoneDistance": "A String", # The audio type that most closely describes the audio being recognized.

434

"interactionType": "A String", # The use case most closely describing the audio content to be recognized.

435

},

436

"maxAlternatives": 42, # Maximum number of recognition hypotheses to be returned.

437

# Specifically, the maximum number of `SpeechRecognitionAlternative` messages

438

# within each `SpeechRecognitionResult`.

439

# The server may return fewer than `max_alternatives`.

440

# Valid values are `0`-`30`. A value of `0` or `1` will return a maximum of

441

# one. If omitted, will return a maximum of one.

442

"audioChannelCount": 42, # The number of channels in the input audio data.

443

# ONLY set this for MULTI-CHANNEL recognition.

444

# Valid values for LINEAR16 and FLAC are `1`-`8`.

445

# Valid values for OGG_OPUS are '1'-'254'.

446

# Valid value for MULAW, AMR, AMR_WB and SPEEX_WITH_HEADER_BYTE is only `1`.

447

# If `0` or omitted, defaults to one channel (mono).

448

# Note: We only recognize the first channel by default.

449

# To perform independent recognition on each channel set

450

# `enable_separate_recognition_per_channel` to 'true'.

Bu Sun Kim

2020-05-27 12:20:54 -0700

[diff] [blame]

451

"encoding": "A String", # Encoding of audio data sent in all `RecognitionAudio` messages.

452

# This field is optional for `FLAC` and `WAV` audio files and required

453

# for all other audio formats. For details, see AudioEncoding.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

454

"model": "A String", # Which model to select for the given request. Select the model

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

455

# best suited to your domain to get best results. If a model is not

456

# explicitly specified, then we auto-select a model based on the parameters

457

# in the RecognitionConfig.

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

458

# <table>

459

# <tr>

460

# <td><b>Model</b></td>

461

# <td><b>Description</b></td>

462

# </tr>

463

# <tr>

464

# <td><code>command_and_search</code></td>

465

# <td>Best for short queries such as voice commands or voice search.</td>

466

# </tr>

467

# <tr>

468

# <td><code>phone_call</code></td>

469

# <td>Best for audio that originated from a phone call (typically

470

# recorded at an 8khz sampling rate).</td>

471

# </tr>

472

# <tr>

473

# <td><code>video</code></td>

474

# <td>Best for audio that originated from from video or includes multiple

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

475

# speakers. Ideally the audio is recorded at a 16khz or greater

476

# sampling rate. This is a premium model that costs more than the

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

477

# standard rate.</td>

478

# </tr>

479

# <tr>

480

# <td><code>default</code></td>

481

# <td>Best for audio that is not one of the specific audio models.

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

482

# For example, long-form audio. Ideally the audio is high-fidelity,

Dan O'Meara

2020-05-01 07:42:23 -0700

[diff] [blame]

483

# recorded at a 16khz or greater sampling rate.</td>

484

# </tr>

485

# </table>

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

486

},

487

"audio": { # Contains audio data in the encoding specified in the `RecognitionConfig`. # Required. The audio data to be recognized.

488

# Either `content` or `uri` must be supplied. Supplying both or neither

489

# returns google.rpc.Code.INVALID_ARGUMENT. See

490

# [content limits](https://cloud.google.com/speech-to-text/quotas#content).

491

"uri": "A String", # URI that points to a file that contains audio data bytes as specified in

492

# `RecognitionConfig`. The file must not be compressed (for example, gzip).

493

# Currently, only Google Cloud Storage URIs are

494

# supported, which must be specified in the following format:

495

# `gs://bucket_name/object_name` (other URI formats return

496

# google.rpc.Code.INVALID_ARGUMENT). For more information, see

497

# [Request URIs](https://cloud.google.com/storage/docs/reference-uris).

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

498

"content": "A String", # The audio data bytes encoded as specified in

499

# `RecognitionConfig`. Note: as with all bytes fields, proto buffers use a

500

# pure binary representation, whereas JSON representations use base64.

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

},

}

x__xgafv: string, V1 error format.

Allowed values

1 - v1 error format

2 - v2 error format

Returns:

An object of the form:

511

512

{ # The only message returned to the client by the `Recognize` method. It

513

# contains the result as zero or more sequential `SpeechRecognitionResult`

514

# messages.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

515

"results": [ # Sequential list of transcription results corresponding to

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

516

# sequential portions of audio.

517

{ # A speech recognition result corresponding to a portion of the audio.

Bu Sun Kim

2020-05-27 12:20:54 -0700

[diff] [blame]

518

"channelTag": 42, # For multi-channel audio, this is the channel number corresponding to the

519

# recognized result for the audio from that channel.

520

# For audio_channel_count = N, its output values can range from '1' to 'N'.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

521

"alternatives": [ # May contain one or more recognition hypotheses (up to the

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

522

# maximum specified in `max_alternatives`).

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

523

# These alternatives are ordered in terms of accuracy, with the top (first)

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

524

# alternative being the most probable, as ranked by the recognizer.

525

{ # Alternative hypotheses (a.k.a. n-best list).

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

526

"transcript": "A String", # Transcript text representing the words that the user spoke.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

527

"confidence": 3.14, # The confidence estimate between 0.0 and 1.0. A higher number

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

528

# indicates an estimated greater likelihood that the recognized words are

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

529

# correct. This field is set only for the top alternative of a non-streaming

530

# result or, of a streaming result where `is_final=true`.

531

# This field is not guaranteed to be accurate and users should not rely on it

532

# to be always provided.

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

533

# The default of 0.0 is a sentinel value indicating `confidence` was not set.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

534

"words": [ # A list of word-specific information for each recognized word.

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

535

# Note: When `enable_speaker_diarization` is true, you will see all the words

536

# from the beginning of the audio.

537

{ # Word-specific information for recognized words.

Bu Sun Kim

2020-07-22 17:02:09 -0700

[diff] [blame^]

538

"speakerTag": 42, # Output only. A distinct integer value is assigned for every speaker within

539

# the audio. This field specifies which one of those speakers was detected to

540

# have spoken this word. Value ranges from '1' to diarization_speaker_count.

541

# speaker_tag is set if enable_speaker_diarization = 'true' and only in the

542

# top alternative.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

543

"endTime": "A String", # Time offset relative to the beginning of the audio,

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

544

# and corresponding to the end of the spoken word.

545

# This field is only set if `enable_word_time_offsets=true` and only

546

# in the top hypothesis.

547

# This is an experimental feature and the accuracy of the time offset can

548

# vary.

Bu Sun Kim

2020-05-20 12:08:20 -0700

[diff] [blame]

549

"startTime": "A String", # Time offset relative to the beginning of the audio,

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

550

# and corresponding to the start of the spoken word.

551

# This field is only set if `enable_word_time_offsets=true` and only

552

# in the top hypothesis.

553

# This is an experimental feature and the accuracy of the time offset can

554

# vary.

Bu Sun Kim

2020-05-27 12:20:54 -0700

[diff] [blame]

555

"word": "A String", # The word corresponding to this set of information.

Bu Sun Kim

2019-06-14 16:50:42 -0700

[diff] [blame]

556

},

557

],

Sai Cheemalapati

2017-06-06 18:46:08 -0400

[diff] [blame]

558

},

559

],

Sai Cheemalapati