- Fix unsafe convertion of
str
to*const c_char
.
- Mark Sync for thread share of resource handle.
- Add synthesis audio format configuration.
- Add
synth_voice_name
to RecognizerConfig. - Change
voice_name
totrans_voice_name
for translator voice name.
- Wrap synthesis functions and increase minor version number.
- Add Text-To-Speech sample and command in Makefile.
DeriveHandle
release function is optional now.- Module
audio
is refactored with many new audio stream types.
- Public export
properties
module, open more control to application.
- Assume that the audio stream should be quantized. Sacrifice efficiency to fix the issue of "invalid audio format".
- Ensure connection of session is closed when recognizer is dropping.
- Ensure session of recognizer is stopped.
- Fix the potential memory leak if Speech SDK does not call
on_stream_close
.
- Simplify the
error
module. - Fix bug of
Box::from_raw
ofStreamReader
. - Fix bug of
pull audio input stream
close function. - Replace member stream of Recognizer to audio with different type.
- Add
update
command to Makefile for upgrading of dependencies. - Add
get_sdk
command to Makefile to fetch latest version of SpeechSDK for running ofasr_simple
. - Add more conversions for std error types to
SpxError
- Add
pull mode
of streaming.
- Fetch & extract
SpeechSDK
inbuild.rs
.
- Remove SpeechSDK, download that on demand.
- Exclude
asr_simple.rs
for leak of keys. - Exclude
SpeechSDK
to reduce size of crate.
- Support microphone input of audio stream.
- Upgrade Microsoft Speech SDK to v1.5.0.
- Add macOS support.
- Add
endpoint
property inRecognizerConfig
for customized model.
- Fix copy/paste bug of
RecognitionResult::offset
.
Recognizer::recognize
return fullEventResult
now.
- Add section for translation and synthesis in
README.md
.
- Translation and translator synthesizing audio.
- More comments copied from Microsoft Speech SDK.
- Event of
Canceled
cause stop of recognition. EventsSessionStopped
andDisconnected
will not be triggered.
- Increase the version number for International Women's Day 2019.
- Fix spell and date error.
- Add more content to README.md file.
- Suppress error result of
RecognitionResult::details
. - Change method name of
EventStream::filter
toEventStream::set_filter
for name conflic ofStream::filter
.
- Convert detailed result of intent recognition to
serde_json::Value
type. - Make
CancellationResult::cancellation_error
be generic. - Expose all fields of struct
Recognition
to public.
- Implicitly add
Flags::Session | Flags::Canceld
for Recognizer startup to avoid unresolved future of EventStream.
- Re-export sub-modules of speech.
- Get rid of the warning of cargo publish by moving bindings.rs to output directory.
- Change the crate category to valid "external-ffi-bindings".
- Add properties bag for AudioInput (SpeechSDK v1.3.1).
- Ignore modification of "asr_simple.rs" for risk of keys leak.
- Re-add "bindings.rs" to lock version in phase of development.
- Add more fields in manifest file.
- Fix errors of inner line doc comments.
- Documentation and manifest for publishing.
- Change mod name from
asr
tospeech
.
- Remove
rustc-hash
crate to wait for new version of HashMap. - Update SpeechSDK to version 1.3.1
- Makefile
run
command addLD_LIBRARY_PATH
environment avoid lib version conflication. - Add intent recognization function.
- Change
reason
ofEventResult
toFlags
type. - Change some weird constant types to readable enum types.
- Do not deglob import of speech_api! (match for unimported consts have potential bugs).
- Add Makefile and CHANGELOG.md
- Fix bug of EventStream filter.
- Rename
EventStream::into_json
tojson
- Change
EventStream::once
name and behavior. - Add push stream support.
- Merge Builder into RecognizerConfig.