Qwen3ASR<T>
Models & Types · AiDotNet.SpeechRecognition.AlibabaASR
Qwen3-ASR-1.7B (Qwen Team, 2026): an LLM-based recognizer of the AuT audio encoder and the Qwen3-1.7B decoder, which names the language and then writes the transcript.
Inheritance: NeuralNetworkBase<T> → AudioNeuralNetworkBase<T> → Qwen3ASR<T>
Implements: ISpeechRecognizer<T>
For Beginners
Qwen3-ASR-1.7B turns speech into a few vectors per second with its audio encoder and lets the Qwen3-1.7B language model answer a chat prompt that contains them: it writes “language English”, a marker, and then the transcript. You can force the language, and put hotwords or background text in the prompt as context.
How It Works
The report (§2) builds Qwen3-ASR-1.7B from Qwen3-1.7B, a projector and an AuT encoder (300M parameters, width 1,024) over 128-band Fbank features at a
12.5 Hz token rate. The release’s modeling_qwen3_asr.py fixes the encoder: 100-frame chunks of the Whisper
log-mel spectrogram, three 3×3 stride-2 GELU convolutions (480 channels), a bias-free linear map to the width with
sinusoidal positions restarting every chunk, pre-norm Transformer layers attending within windows of eight chunks, a
layer norm, and the GELU projector to the decoder’s width. The decoder reads the projector’s output in place of the
audio pads in the release’s chat template (system context; user audio between <|audio_start|> and
<|audio_end|>; assistant), and answers language {name}<asr_text>{transcript}; a forced
language moves its prefix into the prompt. Its interleaved multimodal rotary embedding gives audio and text the same
position on all three axes, which is the ordinary rotary embedding. Training is the release’s supervised fine-tuning
of every parameter on the cross-entropy of the answer and <|im_end|> after the prompt.
Example
var architecture = new NeuralNetworkArchitecture<float>(
inputType: InputType.OneDimensional,
taskType: NeuralNetworkTaskType.SpeechRecognition,
inputSize: 160000, outputSize: 151936);
var model = new Qwen3ASR<float>(architecture);
var result = model.Transcribe(waveform);
Constructors
| Constructor | Summary |
|---|---|
Qwen3ASR(NeuralNetworkArchitecture<T> architecture, Qwen3ASROptions options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer) |
|
Qwen3ASR(NeuralNetworkArchitecture<T> architecture, string modelPath, Qwen3ASROptions options) |
Properties
| Property | Summary |
|---|---|
IReadOnlyList<string> SupportedLanguages { get; } |
|
bool SupportsStreaming { get; } |
|
bool SupportsWordTimestamps { get; } |
Methods
| Method | Summary |
|---|---|
IReadOnlyDictionary<string, T> DetectLanguageProbabilities(Tensor<T> audio) |
|
IStreamingTranscriptionSession<T> StartStreamingSession(string language) |
|
ModelMetadata<T> GetModelMetadata() |
|
Task<TranscriptionResult<T>> TranscribeAsync(Tensor<T> audio, string language, bool? includeTimestamps, CancellationToken cancellationToken) |
|
Tensor<T> ForwardForTraining(Tensor<T> input) |
|
TranscriptionResult<T> Transcribe(Tensor<T> audio, string language, bool? includeTimestamps) |
Transcribes audio: the decoder’s greedy answer, split at <asr_text> into the language it names and the transcript. |
string DetectLanguage(Tensor<T> audio) |
The language the model names before the transcript. |