Repository navigation
Add Paraformer Rust API example add rust paraformer example - #3707
Conversation
dosubot
Bot
added
the
size:L
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughAdds an offline Paraformer speech-recognition Rust example, a model-download runner script, README usage documentation, and Rust test-script integration with cleanup. ChangesParaformer Rust Example
Estimated code review effort: 3 (Moderate) | ~20 minutes Sequence Diagram(s)sequenceDiagram
participant RunScript
participant ParaformerExample
participant OfflineRecognizer
RunScript->>ParaformerExample: pass WAV, model, and tokens paths
ParaformerExample->>OfflineRecognizer: create configured recognizer
ParaformerExample->>OfflineRecognizer: accept waveform and decode
OfflineRecognizer-->>ParaformerExample: return recognition result
Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a new Rust example (paraformer.rs) and an accompanying runner script (run-paraformer.sh) to demonstrate offline speech recognition using the Paraformer model with the sherpa-onnx Rust API. The feedback highlights a few key improvements: updating the main README.md to document this new example, handling a potential division-by-zero error when calculating the Real-Time Factor (RTF) with zero-duration audio, and improving the portability of the download script by supporting both wget and curl.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
| // This file demonstrates how to use non-streaming Paraformer with | ||
| // sherpa-onnx's Rust API for offline speech recognition. | ||
| // | ||
| // See ../README.md for how to run it. |
There was a problem hiding this comment.
The comment mentions See ../README.md for how to run it., but rust-api-examples/README.md has not been updated in this pull request to include the new Paraformer example. Please update rust-api-examples/README.md to add this example to the table of examples and the 'Run it' section.
| println!("Decoded text: {}", result.text); | ||
|
|
||
| let total_time = creation_elapsed + recognition_elapsed; | ||
| let rtf = recognition_elapsed / audio_duration; |
There was a problem hiding this comment.
If audio_duration is 0.0 (e.g., if the WAV file contains no samples or has an invalid sample rate of 0), dividing recognition_elapsed by audio_duration will result in a division by zero, producing NaN or Infinity. Consider adding a guard to handle this case safely.
| let rtf = recognition_elapsed / audio_duration; | |
| let rtf = if audio_duration > 0.0 { | |
| recognition_elapsed / audio_duration | |
| } else { | |
| 0.0 | |
| }; |
| model=sherpa-onnx-paraformer-zh-small-2024-03-09 | ||
|
|
||
| if [ ! -d $model ]; then | ||
| wget $repo_url/$model.tar.bz2 |
There was a problem hiding this comment.
Using only wget can cause the script to fail on systems where wget is not installed by default (such as macOS). Consider checking for both wget and curl to make the script more portable.
| wget $repo_url/$model.tar.bz2 | |
| if command -v wget &>/dev/null; then | |
| wget $repo_url/$model.tar.bz2 | |
| elif command -v curl &>/dev/null; then | |
| curl -OL $repo_url/$model.tar.bz2 | |
| else | |
| echo "Error: Neither wget nor curl found. Please install one of them." | |
| exit 1 | |
| fi |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@rust-api-examples/examples/paraformer.rs`:
- Around line 97-99: The failure branch in paraformer.rs only logs the missing
recognition result and then allows the example to exit successfully. Update the
error path in the recognition result handling block so it terminates the process
with a non-zero exit code when decoding returns no result. Use the existing
result-handling logic around the Paraformer recognition flow to ensure the
failure is propagated to run-paraformer.sh.
In `@rust-api-examples/run-paraformer.sh`:
- Around line 8-17: Anchor the run-paraformer.sh script to its own directory
before using any relative paths, since the current $model, wget, tar, and cargo
run references are resolved from the caller’s CWD instead of the script
location. Update the script to determine its directory first and then build all
model, download, and example paths relative to that location so running the
script from anywhere still downloads/extracts into rust-api-examples and passes
correct paths to cargo run. Refer to the existing $model setup and the final
cargo run invocation when making the change.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 61e24793-8aad-4a59-b2ff-4a161dca9eb9
📒 Files selected for processing (2)
rust-api-examples/examples/paraformer.rsrust-api-examples/run-paraformer.sh
| // This file demonstrates how to use non-streaming Paraformer with | ||
| // sherpa-onnx's Rust API for offline speech recognition. | ||
| // | ||
| // See ../README.md for how to run it. |
There was a problem hiding this comment.
../README.md is not updated for your changes. Can you update it?
| @@ -0,0 +1,17 @@ | |||
| #!/usr/bin/env bash | |||
There was a problem hiding this comment.
Can you also update
https://github.com/k2-fsa/sherpa-onnx/blob/master/.github/scripts/test-rust.sh
to test your changes?
|
Could you fix the comments? |
|
I checked the review comments against the current head and they still look actionable. A small update should be enough to unblock this PR:
For the Rust file, the smallest style-compatible change is probably: if audio_duration <= 0.0 {
eprintln!("WAV file has no samples");
std::process::exit(1);
}
// ... after decode
let Some(result) = stream.get_result() else {
eprintln!("Failed to get recognition result");
std::process::exit(1);
};
println!("Decoded text: {}", result.text);For the runner script, this pattern would address the CWD and downloader comments: #!/usr/bin/env bash
set -ex
script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
cd "$script_dir"
repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models
model=sherpa-onnx-paraformer-zh-small-2024-03-09
archive="$model.tar.bz2"
if [ ! -d "$model" ]; then
if command -v wget >/dev/null 2>&1; then
wget -O "$archive" "$repo_url/$archive"
elif command -v curl >/dev/null 2>&1; then
curl -L -o "$archive" "$repo_url/$archive"
else
echo "Please install wget or curl" >&2
exit 1
fi
tar xvf "$archive"
rm "$archive"
fi
cargo run --example paraformer -- --model "./$model/model.int8.onnx" --tokens "./$model/tokens.txt" --wav "./$model/test_wavs/0.wav"README placement: add ./run-paraformer.shThat should cover the maintainer's "fix the comments" request without changing the example's core API usage. |
|
Thanks for the review. I updated the Rust API README to include the Paraformer example, added a guard for zero-duration audio before calculating RTF, made the no-result path exit with a non-zero status, updated run-paraformer.sh to resolve paths from the script directory while supporting both wget and curl, and added run-paraformer.sh to .github/scripts/test-rust.sh. And I fixed the comment. |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@rust-api-examples/README.md`:
- Line 94: Add a blank line immediately after the examples table entry ending
with the paraformer row and before the following README content, resolving the
markdownlint MD058 violation without changing the table content.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 61219ce4-3dba-41e9-8b89-4998c6f3c2ce
📒 Files selected for processing (4)
.github/scripts/test-rust.shrust-api-examples/README.mdrust-api-examples/examples/paraformer.rsrust-api-examples/run-paraformer.sh
🚧 Files skipped from review as they are similar to previous changes (2)
- rust-api-examples/run-paraformer.sh
- rust-api-examples/examples/paraformer.rs
|
Can you fix the merge conflicts? |
ac045c4 to
ed767b9
Compare
|
Thanks, I rebased my branch on the latest upstream master and resolved the README merge conflict. I kept the new upstream README entries and updated the Paraformer entry/section numbering so it does not conflict with the newer examples. |
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@rust-api-examples/README.md`:
- Line 96: Update the Paraformer index row to link to the Example 51 heading
anchor, replacing the incorrect example-49 anchor while preserving the existing
label and description.
- Around line 94-97: Resolve the merge conflict in the README Examples table by
removing the literal conflict markers and retaining the intended Paraformer
entry as a valid Markdown table row. Ensure the surrounding table remains
correctly formatted.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 174def7b-975a-4758-ae44-fd99acc5847b
📒 Files selected for processing (4)
.github/scripts/test-rust.shrust-api-examples/README.mdrust-api-examples/examples/paraformer.rsrust-api-examples/run-paraformer.sh
🚧 Files skipped from review as they are similar to previous changes (3)
- .github/scripts/test-rust.sh
- rust-api-examples/run-paraformer.sh
- rust-api-examples/examples/paraformer.rs
|
Thanks! |
3 participants
What does this PR do?
This PR adds a non-streaming Paraformer example for the Rust API in
sherpa-onnx. The example shows how to load the Paraformer model, tokens file, and a WAV file, then run offline speech recognition and print the decoded text. I also added a small performance summary so users can see the audio duration, recognition time, and real-time factor.Why was this PR needed?
The project already had a Paraformer example for the C API, but I noticed there was not a matching non-streaming Paraformer example in the Rust API examples folder. Since issue #3210 asks for more Rust API examples, I focused on the Paraformer part and used
c-api-examples/paraformer-c-api.cas my main reference.This should make it easier for Rust users to understand how to run a Paraformer ASR model with
sherpa-onnx.Relevant issue
Addresses part of #3210.
What changed?
rust-api-examples/examples/paraformer.rsrust-api-examples/run-paraformer.shrust-api-examples/README.mdto include the new Paraformer examplerun-paraformer.shto.github/scripts/test-rust.shwgetandcurlTesting
I tested that the new example compiles with:
I also tested it locally with the Paraformer model package using:
The example ran successfully and produced decoded text along with the performance summary.
Checklist
cargo check --example paraformerSummary by CodeRabbit