Protocol Documentation

The license token acquired from the license server.

The language code.

The english name of the language.

The native name of the language.

The writing direction.

The language details.

The domain.

List of available languages for ILS.

The maximum number of character per subtitle line.

The maximum numbers of lines in a subtitle block.

The maximum duration of a subtitle block in seconds.

The minimum duration of a subtitle block in seconds. The minimum duration of a subtitle block in seconds.

The minimum spacing between two subsequent subtitle blocks in seconds

Split sentences at gaps larger than this duration (in seconds).

E.g. '>>' or '-'. Will be added in case of speaker changes when input format is XML and speaker labels are available. Default is to not add a symbol.

Gaps shorter than this duration (in seconds) will be closed to create back-to-back subtitles by increasing the first subtitle's end time.

The maximum number of characters per second in a subtitle block (soft constraint). In the ASR use case, achieved by increasing end time of subtitle blocks. For translation, where timings are not changed, the constraint influences the actual segmentation instead, the number of allowed characters per block will be accounted for (if possible).

Enforces a specific behavior of when to put subsequent sentences onto the same line. If set to true, put short sentences into the same line as preceding or following sentence wherever possible (see also 'multi_sentence_lines_max_pause'). If set to false, keep all sentences separate, this may however lead to short blocks that violate the minimum duration in some cases. If not set, no specific behavior is enforced and the algorithm tries to make optimal decisions per individual case, possibly depending on other configuration fields and language-specific defaults.

Enforces a specific behavior of when to put several sentences into one block. If set to true, put multiple sentences into one block separated by line breaks wherever possible (see also 'multi_sentence_blocks_max_pause'), if spoken by the same speaker. If set to false, start each sentence in a new block. If not set, no specific behavior is enforced and the algorithm tries to make optimal decisions per individual case, possibly depending on other configuration fields and language-specific defaults. If no speaker ids are available (via speaker diarization), all sentences are assumed to be spoken by the same speaker.

If set, implies 'use_multi_sentence_blocks' but allow putting multiple sentences into the same block even if spoken by different speakers. See also the 'dialogue_dash' option.

Only allow putting multiple sentences into one line according to 'use_multi_sentence_lines' if the pause between the sentences (in seconds) is not longer than this value.

Only allow putting multiple sentences into one block according to 'use_multi_sentence_blocks' if the pause between the sentences (in seconds) is not longer than this value.

If 'use_multi_speaker_blocks' is set, use the following string to indicate speaker changes within multi-speaker blocks. Spacing sensitive, for example set to "- " to use hyphen with space. Use empty string to not add a symbol at all. Differs from 'new_speaker_symbol' in that the symbol is not added at every speaker change, but only where necessary to distinguish speakers within a multi-speaker block.

The license token acquired from the license server.

The format of the source document.

The source document.

The language code of the source document.

The optional parameters to tune the output.

The optional domain name (leave empty if in doubt).

The target document as SRT.

A structured response of the subtitles.

The index of the subtitle.

The start time of the subtitle.

The stop time of the subtitle.

All lines of the subtitle.

The line of text.

The speaker id if available.

All subtitles.

The sample type.

The sample rate in Hz

The number of channels (only mono supported at this time).

The Domain name.

The language details.

True if this model supports speaker diarization.

The List of all available recognition engines.

Each speaker in an audio is labeled with an unique id.

Some diarization models provide the name of pre-defined speakers.

The stable transcription from the recognizer.

The stable word after punctuation and capitalization.

True if the text is supposed to be attached to the previous text without whitespace, false otherwise.

True if the text is a sentence end marker, false otherwise.

recognizer timestamp

The start time of the segment (start of the first word).

The stop time of the segment (stop of the last word).

All stable timed transcription details of this segment.

The full text post-processed by ITN (inverse text normalization).

Segmentation is done at a sentence end. For longer sentences a comma might trigger a sentence end as well to ensure lower latencies in a real-time subtitling scenario. This parameter sets the minimum number of tokens (words) in a new sentence that must be seen before a comma can trigger a sentence end. If not set it is set to the models default value.

The domain name (leave empty if not sure).

The language code of the audio.

The license token acquired from the license server.

The audio configuration.

The speech configuration.

The diarizer configuration.

Translate configurations for each requests target language.

The segmenter configuration.

The Configuration message.

PCM maudio data matching sample_type/sample_rate/channels in Recognize2AudioConfig.

A progress status message.

A transcription result from the recognizer.

A transcription result after being punctuated and capitalized.

The transcription of a segment (sentence) after being processed by ITN.

A translation of a segment.

The start time of a word.

The stop time of a word.

The recognized word.

The confidence score between 0.0 and 1.0

For multi-language models this contains the language code of the word.

List of still unstable transcriptions.

List of stable transcriptions.

The Domain name (leave empty if not sure).

The language code of the requested target language.

Optional use of a glossary.

Optional meta option settings.

The start time of the translated text segment.

The stop time of the translated text segment.

The translated text.

The domain name of the selected translation model.

The language code of the translated text.

The encoding of the audio data.

The sample rate of the audio data in Hz.

language code of the audio

The domain name (leave empty if not sure).

The maximum mutable suffix length (leave empty if not sure).

Enable diarization.

Load custom vocabulary with this id.

The domain name.

The audio sample rate in Hz.

The language details.

The list of available direct translations.

The value is true if and only if diarization is available, false otherwise.

The list of all available recognition engines.

The segment details.

List of all speaker sections.

The speaker name.

he transcribed text.

The postprocessed text.

The translated test.

The start time of the speaker section in milliseconds.

The stop time of the speaker section in milliseconds.

The segment details.

The source text.

The postprocessed version of the source text.

The audio decoder progress time in seconds.

The segmenter progress time in seconds.

The recognizer progress time in seconds.

The remaining time in seconds.

The unique segment id.

Complete is true if and only if this is the last message for this response type with the given id, false otherwise.

The unique segment id.

The progress time in seconds since the last segment end in seconds.

The Start time of the speaker in milliseconds.

The speaker name.

The license token acquired from the license server.

The audio data configuration.

The optional translate configuration.

The configuration message. This must be the first message.

The audio data.

A transcription response.

A segment end response.

A progress status response.

A postprocessing response.

A translation response.

A speaker change response.

A diarization response.

The segment details.

The transcribed segment in one string.

The list of all transcribed words with detailed information.

The interval in which translation results are produced.

The lang code of the requested target language.

Enable use of a registered glossary.

The segment details.

The source text for this translation.

The translated text

The recognized word.

The start time in milliseconds.

The stop time in milliseconds.

The confidence score (between 0.0 and 1.0).

The license token acquired from the license server.

The speak lexicon.

The unique id of the lexicon.

The expiration date of the lexicon in seconds since epoch.

The license token acquired from the license server.

The encoding of the reference audio data.

The reference audio data.

The unique id of the reference audio data.

The expiration time in second since epoch.

The domain name.

The language details.

The speaker name.

If true, the TTS allows to adapt the speaker voice with reference audio data.

The Gender of the TTS speaker (might be empty, e. g. for adaptable speakers).

List of all available TTS engines.

The license token acquired from the license server.

The word or a small phrase.

The tts language code.

The speaker to generate the audio.

The domain name

The optional language code, e. g. for pronouncing foreign words.

The symbols.

The chunk of audio data as a wave file.

Maps source word/phrase to replace rule.

If true, the pattern matching will be performed case sensitive.

The desired tts input as text.

The desired tts input as symbols.

The encoding of the TTS audio data.

The license token acquired from the license server.

The unique id of the lexicon.

The license token acquired from the license server.

The unique id of the reference audio data.

The license token acquired from the license server.

The requested encoding of the output audio stream.

The input text.

The language code of the input text.

The TTS speaker name.

The domain name.

Tempo specifies the playback speed of the generated audio file. If set to 0.0 it defaults internally to normal speed (1.0), values between 0.5 and 1.5 are applicable.

The unique id of a registered reference audio to adapt the speakers voice.

The unique id of a register lexicon.

The first and only the first message is a meta message describing the content of the following audio messages.

The audio containing the generated speech.

The domain

The language details.

The transformation mode.

List of available ITN languages.

The license token acquired from the license server.

The source text.

The source language code.

The domain.

The layout of the text for plain source texts.

The targe text.

The license token acquired from the license server.

The glossary.

The unique id of the glossary.

Minimum expiration time of the glossary.

The domain name.

The language details for the source text.

The language details for the target text.

The available source text formats.

The available gender codes.

The available genre codes.

The available length codes.

The available style codes.

The available topic codes.

True iff model supports translation with extended context.

True iff model supports computation of confidence scores.

True iff use of glossary for source language is supported.

True iff annotations with translation, but without tag are supported.

True iff annotations with both translation and tag are supported.

True iff annotations with tag, but without translation are supported.

True iff subtitle condensation is available.

List of all available translation engines.

The source language code.

The target language code.

List of all suggested translations.

The source word.

The suggested target word

The optional alternative suggested translations.

The unique id of the registered glossary.

Numbers can be are tagged in order to show which number on the source side has been used in which position on the target side e.g.: Input: “Tom bought 6 eggs and 4 tomatoes.” Annotation: “Tom bought <1>6</1> eggs and <2>4</2> tomatoes.” Translation: “Tom hat <1>6</1> Eier und <2>4</2> Tomaten gekauft.”

The license token acquired from the license server.

The unique id of the glossary.

The maximum number of character per subtitle line.

The maximum numbers of lines in a subtitle block.

The maximum number of characters per second in a subtitle block. If set, translation will be condensed to better meet this limit.

Split sentences at gaps larger than this duration.

The gender_code controls the grammar/morphological forms of the translation (for example, female vs. male 1st person verb forms in Slavic languages), in dependency on the gender of the speaker/author of the sentence.

The genre_code provides information about the genre of the input text, the translation is then adjusted to fit a particular genre (for example, news, patents, etc.).

The length_code lets you control the length of the translation output, in relation to the length of the input sentence.

The style_code controls the style/formality of the translation (for example, using formal "you" instead of informal "you").

The topic_code provides information about the topic of the input text, the translation is then adjusted to fit a particular topic (for example, politics, sports, culture within the genre of "news").

The license token acquired from the license server.

The language code of your source (input) text.

The language code of your desired target language.

The domain name of the desired translation model.

The source text in form of formatted subtitles.

The global meta options to tune the translation.

The line segmentation options for formatting the subtitles.

Enable use of a registered glossary.

Compute the confidence score per subtitle.

Use extended context during translation.

Optimize translation to better adhere to maximum_lines line segmentation option.

The translated subtitles.

The detailed results for each translated segment.

The unique index of the subtitle.

The start time of the subtitle.

The stop time of the subtitle.

The lines of text of the subtitle.

This flag can be used to update parts of an earlier translation. The output will contain translations only of the subtitles with this flag set, as well as possibly surrounding subtitles that are affected by the updated translation. The client can assemble the updated subtitle file via the subtitle indices.

One line of a subtitle block. May include formatting tags, e.g. '<i> ... </i>'. In addition, we support tags for glossary term translation of the form '<term>glossary term</term>' or '<term target="Glossareintrag">glossary term</term>'. The 'target' attribute defines the desired translation of the term between the tags, and a missing 'target' attribute means the term should be kept as is during translation.

Use extended context during translation, overrides global setting if set for this line.

Overrides global meta options if set for this line.

The unique index of the subtitle

The start time of the subtitle.

The stop time of the subtitle.

The lines of text of the subtitle.

The confidence score if requested.

The line of text.

Index of first word in the word_items list.

Index of the last word (exclusive).

The tag of the corresponding source annotation (if it was set) to identify the span.tag in the translation.

Optional alternative translations.

The tart tag to be inserted before the span when converting to plain text representation, e.g. <i>.

The stop tags to be inserted after the span when converting to plain text representation, e.g. </i>.

The translated text.

The confidence score for this hypothesis.

List of all word items.

List of all target annotations.

The gender_code controls the grammar/morphological forms of the translation (for example, female vs. male 1st person verb forms in Slavic languages), in dependency on the gender of the speaker/author of the sentence.

The Genre_code provides information about the genre of the input text, the translation is then adjusted to fit a particular genre (for example, news, patents, etc.).

The length_code lets you control the length of the translation output, in relation to the length of the input sentence.

The style_code controls the style/formality of the translation (for example, using formal "you" instead of informal "you").

The topic_code provides information about the topic of the input text, the translation is then adjusted to fit a particular topic (for example, politics, sports, culture within the genre of "news").

The license token acquired from the license server.

The plain source text.

The language code of the provided source text.

The language code of the requested target language.

The domain name.

The source text format.

The optional parameters if source text is a SRT.

The optional meta parameters to tune the translation.

Enable use of extended context.

Enable computation confidence scores, this should only be enabled when required, since it slows down the translation.

The maximum number of hypotheses to be returned (minimum and default if not set is 1).

Enable use of a glossary.

Enable parsing tags of the form '<tag>tagged term</tag>', see above.

Optimize translation to better adhere to maximum_lines line segmentation option.

The layout of the text for plain source texts.

Target text of the best hypotheses in one string.

Detailed results for each translated segment.

List of all hypotheses.

List of all source annotations.

List of all source words matching the indices in the annotations.

The annotation.

The generated target.

The word.

True if word is attached to the previous word on the left, false otherwise.

The maximum number of characters per line.

The maximum number of lines per frame.

The maximum number of characters per second in a subtitle block. If set, translation will be condensed to better meet this limit.

Split sentences at gaps larger than this duration.

The word.

The words confidence score

True if word is attached to the previous word on the left, false otherwise.

Deprecated. Replaced by tag in the annotation message.