Blame - docs/dyn/speech_v1p1beta1.speech.html - platform/external/python/google-api-python-client

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

[diff] [blame]

100

# process the request.

101

# request.

Bu Sun Kim

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

[diff] [blame]

102

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

103

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

104

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

105

# `false`.

106

"alternativeLanguageCodes": [ # A list of up to 3 additional

107

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

108

# listing possible alternative languages of the supplied audio.

109

# See [Language

110

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

111

# of the currently supported language codes. If alternative languages are

112

# listed, recognition result will contain recognition in the most likely

113

# language detected including the main language_code. The recognition result

114

# will include the language tag of the language detected in the audio. Note:

115

# This feature is only supported for Voice Command and Voice Search use cases

116

# and performance may vary for other use cases (e.g., phone call

# transcription).

"A String",

],

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

121

# parameters to make diarization better suited for your application.

122

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

123

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

124

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

125

# identify the speakers in the conversation over time.

126

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

127

# in the top alternative of the FINAL SpeechRecognitionResult.

128

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

129

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

130

# in the WordInfo.

131

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

132

# flexibility by allowing the system to automatically determine the correct

133

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

134

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

135

# flexibility by allowing the system to automatically determine the correct

136

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

137

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

138

},

139

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

140

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

141

# Example: "en-US".

142

# See [Language

143

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

144

# of the currently supported language codes.

145

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

146

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

147

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

148

# won't be filtered out.

149

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

150

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

151

# in the WordInfo.

152

# Note: Use diarization_config instead.

153

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

154

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

155

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

156

# the audio.

157

#

158

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

159

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

160

# of the specified model.

Bu Sun Kim

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

[diff] [blame]

161

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

Bu Sun Kim

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

[diff] [blame]

162

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

163

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

164

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

165

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

166

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

Bu Sun Kim

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

[diff] [blame]

167

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

168

"obfuscatedId": "A String", # Obfuscated (privacy-protected) ID of the user, to identify number of

169

# unique users using the service.

170

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

171

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

172

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

173

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

174

# 'Cardioid Microphone'.

175

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

176

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

177

# A list of possible audio mime types is maintained at

178

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

179

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

180

# hearings from 2012".

Bu Sun Kim

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

[diff] [blame]

181

},

182

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

183

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

184

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

185

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

186

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

187

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

188

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

189

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

Bu Sun Kim

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

[diff] [blame]

190

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

191

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

192

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

193

# request is billed cumulatively for all channels recognized:

194

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

Bu Sun Kim

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

[diff] [blame]

195

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

196

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

197

# requests in other languages has no effect at all.

198

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

199

"adaptation": { # Speech adaptation configuration. # Speech adaptation configuration improves the accuracy of speech

Dan O'Meara

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

[diff] [blame]

200

# recognition. When speech adaptation is set it supersedes the

201

# `speech_contexts` field. For more information, see the [speech

202

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

203

# documentation.

Bu Sun Kim

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

[diff] [blame]

204

"phraseSets": [ # A collection of phrase sets. To specify the hints inline, leave the

205

# phrase set's `name` blank and fill in the rest of its fields. Any

Dan O'Meara

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

[diff] [blame]

206

# phrase set can use any custom class.

Bu Sun Kim

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

[diff] [blame]

207

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

Dan O'Meara

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

[diff] [blame]

208

# in the results.

Bu Sun Kim

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

[diff] [blame]

209

"name": "A String", # The resource name of the phrase set.

210

"phrases": [ # A list of word and phrases.

211

{ # A phrases containing words and phrase "hints" so that

Dan O'Meara

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

[diff] [blame]

212

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

213

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

214

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

215

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

216

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

217

#

218

# List items can also include pre-built or custom classes containing groups

219

# of words that represent common concepts that occur in natural language. For

220

# example, rather than providing a phrase hint for every month of the

Bu Sun Kim

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

[diff] [blame]

221

# year (e.g. "i was born in january", "i was born in febuary", ...), use the

222

# pre-built `$MONTH` class improves the likelihood of correctly transcribing

223

# audio that includes months (e.g. "i was born in $month").

224

# To refer to pre-built classes, use the class' symbol prepended with `$`

225

# e.g. `$MONTH`. To refer to custom classes that were defined inline in the

226

# request, set the class's `custom_class_id` to a string unique to all class

227

# resources and inline classes. Then use the class' id wrapped in $`{...}`

228

# e.g. "${my-months}". To refer to custom classes resources, use the class'

229

# id wrapped in `${}` (e.g. `${my-months}`).

Bu Sun Kim

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

[diff] [blame]

230

"boost": 3.14, # Hint Boost. Overrides the boost set at the phrase set level.

Dan O'Meara

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

[diff] [blame]

231

# Positive value will increase the probability that a specific phrase will

232

# be recognized over other similar sounding phrases. The higher the boost,

233

# the higher the chance of false positive recognition as well. Negative

234

# boost values would correspond to anti-biasing. Anti-biasing is not

235

# enabled, so negative boost will simply be ignored. Though `boost` can

236

# accept a wide range of positive values, most use cases are best served

237

# with values between 0 and 20. We recommend using a binary search approach

238

# to finding the optimal value for your use case. Speech recognition

239

# will skip PhraseSets with a boost value of 0.

Bu Sun Kim

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

[diff] [blame]

240

"value": "A String", # The phrase itself.

Dan O'Meara

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

[diff] [blame]

241

},

242

],

Bu Sun Kim

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

[diff] [blame]

243

"boost": 3.14, # Hint Boost. Positive value will increase the probability that a specific

244

# phrase will be recognized over other similar sounding phrases. The higher

245

# the boost, the higher the chance of false positive recognition as well.

246

# Negative boost values would correspond to anti-biasing. Anti-biasing is not

247

# enabled, so negative boost will simply be ignored. Though `boost` can

248

# accept a wide range of positive values, most use cases are best served with

249

# values between 0 (exclusive) and 20. We recommend using a binary search

250

# approach to finding the optimal value for your use case. Speech recognition

251

# will skip PhraseSets with a boost value of 0.

Dan O'Meara

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

[diff] [blame]

252

},

253

],

Bu Sun Kim

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

[diff] [blame]

254

"customClasses": [ # A collection of custom classes. To specify the classes inline, leave the

255

# class' `name` blank and fill in the rest of its fields, giving it a unique

Dan O'Meara

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

[diff] [blame]

256

# `custom_class_id`. Refer to the inline defined class in phrase hints by its

257

# `custom_class_id`.

258

{ # A set of words or phrases that represents a common concept likely to appear

259

# in your audio, for example a list of passenger ship names. CustomClass items

260

# can be substituted into placeholders that you set in PhraseSet phrases.

Bu Sun Kim

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

[diff] [blame]

261

"items": [ # A collection of class items.

Dan O'Meara

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

[diff] [blame]

262

{ # An item of the class.

Bu Sun Kim

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

[diff] [blame]

263

"value": "A String", # The class item's value.

Dan O'Meara

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

[diff] [blame]

264

},

265

],

Bu Sun Kim

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

[diff] [blame]

266

"name": "A String", # The resource name of the custom class.

267

"customClassId": "A String", # If this custom class is a resource, the custom_class_id is the resource id

268

# of the CustomClass. Case sensitive.

Dan O'Meara

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

[diff] [blame]

269

},

270

],

271

},

Bu Sun Kim

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

[diff] [blame]

272

"diarizationSpeakerCount": 42, # If set, specifies the estimated number of speakers in the conversation.

273

# Defaults to '2'. Ignored unless enable_speaker_diarization is set to true.

274

# Note: Use diarization_config instead.

Bu Sun Kim

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

[diff] [blame]

275

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

276

# Specifically, the maximum number of `SpeechRecognitionAlternative` messages

277

# within each `SpeechRecognitionResult`.

278

# The server may return fewer than `max_alternatives`.

279

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

280

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

Bu Sun Kim

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

[diff] [blame]

281

"speechContexts": [ # Array of SpeechContext.

282

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

283

# information, see

284

# [speech

285

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

286

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

287

# in the results.

288

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

289

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

290

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

291

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

292

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

293

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

294

#

295

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

296

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

297

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

298

# improves the likelihood of correctly transcribing audio that includes

# months.

"A String",

],

"boost": 3.14, # Hint Boost. Positive value will increase the probability that a specific

303

# phrase will be recognized over other similar sounding phrases. The higher

304

# the boost, the higher the chance of false positive recognition as well.

305

# Negative boost values would correspond to anti-biasing. Anti-biasing is not

306

# enabled, so negative boost will simply be ignored. Though `boost` can

307

# accept a wide range of positive values, most use cases are best served with

308

# values between 0 and 20. We recommend using a binary search approach to

309

# finding the optimal value for your use case.

310

},

311

],

Bu Sun Kim

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

[diff] [blame]

312

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

313

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

314

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

Bu Sun Kim

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

[diff] [blame]

315

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

Bu Sun Kim

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

[diff] [blame]

316

# confidence for those words. If `false`, no word-level confidence

317

# information is returned. The default is `false`.

Bu Sun Kim

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

[diff] [blame]

318

"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]

319

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

320

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

321

# in the RecognitionConfig.

Dan O'Meara

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

[diff] [blame]

322

# <table>

323

# <tr>

324

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

325

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

326

# </tr>

327

# <tr>

328

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

329

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

330

# </tr>

331

# <tr>

332

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

333

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

334

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

335

# </tr>

336

# <tr>

337

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

338

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

Bu Sun Kim

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

[diff] [blame]

339

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

340

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

Dan O'Meara

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

[diff] [blame]

341

# standard rate.</td>

342

# </tr>

343

# <tr>

344

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

345

# <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]

346

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

Dan O'Meara

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

[diff] [blame]

347

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

348

# </tr>

349

# </table>

Bu Sun Kim

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

[diff] [blame]

350

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

351

# ONLY set this for MULTI-CHANNEL recognition.

352

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

353

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

354

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

355

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

356

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

357

# To perform independent recognition on each channel set

358

# `enable_separate_recognition_per_channel` to 'true'.

Bu Sun Kim

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

[diff] [blame]

359

},

360

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

361

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

362

# returns google.rpc.Code.INVALID_ARGUMENT. See

363

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

364

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

365

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

366

# pure binary representation, whereas JSON representations use base64.

367

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

368

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

369

# Currently, only Google Cloud Storage URIs are

370

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

371

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

372

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

373

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

Bu Sun Kim

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

[diff] [blame]

},

}

x__xgafv: string, V1 error format.

Allowed values

1 - v1 error format

2 - v2 error format

Returns:

An object of the form:

384

385

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

386

# network API call.

Bu Sun Kim

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

[diff] [blame]

387

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

388

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

389

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

390

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

391

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

392

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

393

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

394

# `TakeSnapshotResponse`.

395

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

396

},

Bu Sun Kim

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

[diff] [blame]

397

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

398

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

399

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

400

"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.

Bu Sun Kim

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

[diff] [blame]

401

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

402

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

403

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

404

#

405

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

406

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

Bu Sun Kim

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

[diff] [blame]

407

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

408

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

Bu Sun Kim

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

[diff] [blame]

409

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

410

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

Bu Sun Kim

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

[diff] [blame]

411

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

Bu Sun Kim

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

[diff] [blame]

412

# message types for APIs to use.

413

{

Bu Sun Kim

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

[diff] [blame]

414

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

Bu Sun Kim

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

[diff] [blame]

415

},

416

],

417

},

Bu Sun Kim

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

[diff] [blame]

418

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

419

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

420

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

421

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

422

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

423

},

424

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

Bu Sun Kim

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

[diff] [blame]

425

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

426

# available.

Bu Sun Kim

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

[diff] [blame]

}</pre>

</div>

Dan O'Meara

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

[diff] [blame]

431

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

Bu Sun Kim

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

[diff] [blame]

432

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

433

has been sent and processed.

434

435

Args:

Dan O'Meara

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

[diff] [blame]

436

body: object, The request body.

Bu Sun Kim

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

[diff] [blame]

437

The object takes the form of:

438

439

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

Bu Sun Kim

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

[diff] [blame]

440

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

Bu Sun Kim

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

[diff] [blame]

441

# process the request.

442

# request.

Bu Sun Kim

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

[diff] [blame]

443

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

444

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

445

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

446

# `false`.

447

"alternativeLanguageCodes": [ # A list of up to 3 additional

448

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

449

# listing possible alternative languages of the supplied audio.

450

# See [Language

451

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

452

# of the currently supported language codes. If alternative languages are

453

# listed, recognition result will contain recognition in the most likely

454

# language detected including the main language_code. The recognition result

455

# will include the language tag of the language detected in the audio. Note:

456

# This feature is only supported for Voice Command and Voice Search use cases

457

# and performance may vary for other use cases (e.g., phone call

# transcription).

"A String",

],

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

462

# parameters to make diarization better suited for your application.

463

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

464

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

465

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

466

# identify the speakers in the conversation over time.

467

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

468

# in the top alternative of the FINAL SpeechRecognitionResult.

469

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

470

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

471

# in the WordInfo.

472

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

473

# flexibility by allowing the system to automatically determine the correct

474

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

475

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

476

# flexibility by allowing the system to automatically determine the correct

477

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

478

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

479

},

480

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

481

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

482

# Example: "en-US".

483

# See [Language

484

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

485

# of the currently supported language codes.

486

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

487

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

488

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

489

# won't be filtered out.

490

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

491

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

492

# in the WordInfo.

493

# Note: Use diarization_config instead.

494

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

495

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

496

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

497

# the audio.

498

#

499

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

500

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

501

# of the specified model.

Bu Sun Kim

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

[diff] [blame]

502

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

Bu Sun Kim

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

[diff] [blame]

503

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

504

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

505

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

506

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

507

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

Bu Sun Kim

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

[diff] [blame]

508

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

509

"obfuscatedId": "A String", # Obfuscated (privacy-protected) ID of the user, to identify number of

510

# unique users using the service.

511

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

512

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

513

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

514

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

515

# 'Cardioid Microphone'.

516

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

517

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

518

# A list of possible audio mime types is maintained at

519

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

520

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

521

# hearings from 2012".

Bu Sun Kim

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

[diff] [blame]

522

},

523

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

524

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

525

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

526

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

527

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

528

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

529

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

530

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

Bu Sun Kim

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

[diff] [blame]

531

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

532

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

533

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

534

# request is billed cumulatively for all channels recognized:

535

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

Bu Sun Kim

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

[diff] [blame]

536

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

537

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

538

# requests in other languages has no effect at all.

539

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

540

"adaptation": { # Speech adaptation configuration. # Speech adaptation configuration improves the accuracy of speech

Dan O'Meara

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

[diff] [blame]

541

# recognition. When speech adaptation is set it supersedes the

542

# `speech_contexts` field. For more information, see the [speech

543

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

544

# documentation.

Bu Sun Kim

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

[diff] [blame]

545

"phraseSets": [ # A collection of phrase sets. To specify the hints inline, leave the

546

# phrase set's `name` blank and fill in the rest of its fields. Any

Dan O'Meara

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

[diff] [blame]

547

# phrase set can use any custom class.

Bu Sun Kim

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

[diff] [blame]

548

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

Dan O'Meara

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

[diff] [blame]

549

# in the results.

Bu Sun Kim

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

[diff] [blame]

550

"name": "A String", # The resource name of the phrase set.

551

"phrases": [ # A list of word and phrases.

552

{ # A phrases containing words and phrase "hints" so that

Dan O'Meara

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

[diff] [blame]

553

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

554

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

555

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

556

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

557

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

558

#

559

# List items can also include pre-built or custom classes containing groups

560

# of words that represent common concepts that occur in natural language. For

561

# example, rather than providing a phrase hint for every month of the

Bu Sun Kim

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

[diff] [blame]

562

# year (e.g. "i was born in january", "i was born in febuary", ...), use the

563

# pre-built `$MONTH` class improves the likelihood of correctly transcribing

564

# audio that includes months (e.g. "i was born in $month").

565

# To refer to pre-built classes, use the class' symbol prepended with `$`

566

# e.g. `$MONTH`. To refer to custom classes that were defined inline in the

567

# request, set the class's `custom_class_id` to a string unique to all class

568

# resources and inline classes. Then use the class' id wrapped in $`{...}`

569

# e.g. "${my-months}". To refer to custom classes resources, use the class'

570

# id wrapped in `${}` (e.g. `${my-months}`).

Bu Sun Kim

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

[diff] [blame]

571

"boost": 3.14, # Hint Boost. Overrides the boost set at the phrase set level.

Dan O'Meara

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

[diff] [blame]

572

# Positive value will increase the probability that a specific phrase will

573

# be recognized over other similar sounding phrases. The higher the boost,

574

# the higher the chance of false positive recognition as well. Negative

575

# boost values would correspond to anti-biasing. Anti-biasing is not

576

# enabled, so negative boost will simply be ignored. Though `boost` can

577

# accept a wide range of positive values, most use cases are best served

578

# with values between 0 and 20. We recommend using a binary search approach

579

# to finding the optimal value for your use case. Speech recognition

580

# will skip PhraseSets with a boost value of 0.

Bu Sun Kim

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

[diff] [blame]

581

"value": "A String", # The phrase itself.

Dan O'Meara

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

[diff] [blame]

582

},

583

],

Bu Sun Kim

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

[diff] [blame]

584

"boost": 3.14, # Hint Boost. Positive value will increase the probability that a specific

585

# phrase will be recognized over other similar sounding phrases. The higher

586

# the boost, the higher the chance of false positive recognition as well.

587

# Negative boost values would correspond to anti-biasing. Anti-biasing is not

588

# enabled, so negative boost will simply be ignored. Though `boost` can

589

# accept a wide range of positive values, most use cases are best served with

590

# values between 0 (exclusive) and 20. We recommend using a binary search

591

# approach to finding the optimal value for your use case. Speech recognition

592

# will skip PhraseSets with a boost value of 0.

Dan O'Meara

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

[diff] [blame]

593

},

594

],

Bu Sun Kim

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

[diff] [blame]

595

"customClasses": [ # A collection of custom classes. To specify the classes inline, leave the

596

# class' `name` blank and fill in the rest of its fields, giving it a unique

Dan O'Meara

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

[diff] [blame]

597

# `custom_class_id`. Refer to the inline defined class in phrase hints by its

598

# `custom_class_id`.

599

{ # A set of words or phrases that represents a common concept likely to appear

600

# in your audio, for example a list of passenger ship names. CustomClass items

601

# can be substituted into placeholders that you set in PhraseSet phrases.

Bu Sun Kim

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

[diff] [blame]

602

"items": [ # A collection of class items.

Dan O'Meara

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

[diff] [blame]

603

{ # An item of the class.

Bu Sun Kim

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

[diff] [blame]

604

"value": "A String", # The class item's value.

Dan O'Meara

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

[diff] [blame]

605

},

606

],

Bu Sun Kim

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

[diff] [blame]

607

"name": "A String", # The resource name of the custom class.

608

"customClassId": "A String", # If this custom class is a resource, the custom_class_id is the resource id

609

# of the CustomClass. Case sensitive.

Dan O'Meara

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

[diff] [blame]

610

},

611

],

612

},

Bu Sun Kim

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

[diff] [blame]

613

"diarizationSpeakerCount": 42, # If set, specifies the estimated number of speakers in the conversation.

614

# Defaults to '2'. Ignored unless enable_speaker_diarization is set to true.

615

# Note: Use diarization_config instead.

Bu Sun Kim

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

[diff] [blame]

616

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

617

# Specifically, the maximum number of `SpeechRecognitionAlternative` messages

618

# within each `SpeechRecognitionResult`.

619

# The server may return fewer than `max_alternatives`.

620

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

621

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

Bu Sun Kim

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

[diff] [blame]

622

"speechContexts": [ # Array of SpeechContext.

623

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

624

# information, see

625

# [speech

626

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

627

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

628

# in the results.

629

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

630

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

631

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

632

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

633

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

634

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

635

#

636

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

637

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

638

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

639

# improves the likelihood of correctly transcribing audio that includes

# months.

"A String",

],

"boost": 3.14, # Hint Boost. Positive value will increase the probability that a specific

644

# phrase will be recognized over other similar sounding phrases. The higher

645

# the boost, the higher the chance of false positive recognition as well.

646

# Negative boost values would correspond to anti-biasing. Anti-biasing is not

647

# enabled, so negative boost will simply be ignored. Though `boost` can

648

# accept a wide range of positive values, most use cases are best served with

649

# values between 0 and 20. We recommend using a binary search approach to

650

# finding the optimal value for your use case.

651

},

652

],

Bu Sun Kim

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

[diff] [blame]

653

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

654

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

655

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

Bu Sun Kim

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

[diff] [blame]

656

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

Bu Sun Kim

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

[diff] [blame]

657

# confidence for those words. If `false`, no word-level confidence

658

# information is returned. The default is `false`.

Bu Sun Kim

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

[diff] [blame]

659

"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]

660

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

661

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

662

# in the RecognitionConfig.

Dan O'Meara

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

[diff] [blame]

663

# <table>

664

# <tr>

665

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

666

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

667

# </tr>

668

# <tr>

669

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

670

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

671

# </tr>

672

# <tr>

673

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

674

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

675

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

676

# </tr>

677

# <tr>

678

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

679

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

Bu Sun Kim

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

[diff] [blame]

680

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

681

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

Dan O'Meara

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

[diff] [blame]

682

# standard rate.</td>

683

# </tr>

684

# <tr>

685

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

686

# <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]

687

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

Dan O'Meara

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

[diff] [blame]

688

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

689

# </tr>

690

# </table>

Bu Sun Kim

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

[diff] [blame]

691

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

692

# ONLY set this for MULTI-CHANNEL recognition.

693

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

694

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

695

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

696

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

697

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

698

# To perform independent recognition on each channel set

699

# `enable_separate_recognition_per_channel` to 'true'.

Bu Sun Kim

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

[diff] [blame]

700

},

701

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

702

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

703

# returns google.rpc.Code.INVALID_ARGUMENT. See

704

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

705

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

706

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

707

# pure binary representation, whereas JSON representations use base64.

708

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

709

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

710

# Currently, only Google Cloud Storage URIs are

711

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

712

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

713

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

714

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

Bu Sun Kim

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

[diff] [blame]

715

},

Bu Sun Kim

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

[diff] [blame]

716

}

717

718

x__xgafv: string, V1 error format.

Allowed values

1 - v1 error format

2 - v2 error format

Returns:

An object of the form:

725

726

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

727

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

728

# messages.

Bu Sun Kim

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

[diff] [blame]

729

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

Bu Sun Kim

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

[diff] [blame]

730

# sequential portions of audio.

731

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

Bu Sun Kim

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

[diff] [blame]

732

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

Dan O'Meara

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

[diff] [blame]

733

# recognized result for the audio from that channel.

Bu Sun Kim

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

[diff] [blame]

734

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

735

"languageCode": "A String", # Output only. The [BCP-47](https://www.rfc-editor.org/rfc/bcp/bcp47.txt) language tag

736

# of the language in this result. This language code was detected to have

737

# the most likelihood of being spoken in the audio.

738

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

Bu Sun Kim

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

[diff] [blame]

739

# maximum specified in `max_alternatives`).

740

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

741

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

742

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

Bu Sun Kim

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

[diff] [blame]

743

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

Bu Sun Kim

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

[diff] [blame]

744

# indicates an estimated greater likelihood that the recognized words are

745

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

746

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

747

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

748

# to be always provided.

749

# 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]

750

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

751

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

Bu Sun Kim

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

[diff] [blame]

752

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

753

# from the beginning of the audio.

754

{ # Word-specific information for recognized words.

Bu Sun Kim

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

[diff] [blame]

755

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

756

# and corresponding to the end of the spoken word.

757

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

758

# in the top hypothesis.

759

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

760

# vary.

761

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

Bu Sun Kim

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

[diff] [blame]

762

# indicates an estimated greater likelihood that the recognized words are

763

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

764

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

765

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

766

# to be always provided.

767

# 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]

768

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

Bu Sun Kim

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

[diff] [blame]

769

# and corresponding to the start of the spoken word.

770

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

771

# in the top hypothesis.

772

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

773

# vary.

Bu Sun Kim

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

[diff] [blame]

774

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

Bu Sun Kim

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

[diff] [blame]

775

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

776

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

777

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

778

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

779

# top alternative.

Bu Sun Kim

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

[diff] [blame]

},

],

},

],

Bu Sun Kim